Demonstration and Research Pest Control Oregon Pesticide Safety Education Program Applicator Training Manual 2 | Demonstration and Research Pest Control 00 About this manual | 3

About this manual

This manual has been designed for research Demonstration and Research Pest Control scientists, Extension agents, Extension category and for meeting ODA’s specific specialists, industry representatives, standards of competency established by 40 employees of federal and state government, CFR 171.4(c)(10), ORS 634 and OAR 603- and other professionals who conduct field 057. In addition to the material covered in research with unregistered experimental the manual, candidates may be responsible pesticides or demonstrations with registered for information contained in one or more of pesticides. It is intended to provide the the category manuals that are appropriate information needed to meet the minimum for a particular pest control activity. standards of the U.S. Environmental Cover photos courtesy of Laurie Gordon, Protection Agency (EPA and the Oregon Oregon Department of Agriculture. Department of Agriculture (ODA) certification of Pesticide Consultants and Commercial and Public applicators in the Demonstration and Research category. These requirements have been established by EPA and ODA through the Federal Insecticide, Fungicide, and Rodenticide Act (FIFRA) (7 U.S. Code 136i), Title 40 Code of Federal Regulations part 171 (40 CFR 171), ORS 634, and OAR 603-057. This material is a supplement to the information contained in the study manuals suggested for the Laws and Safety exam or the Consultant exam. It should not be used to prepare for certification without referring to these manuals. Thorough study of the information in all study manuals will help users prepare for examinations in the 4 | Demonstration and Research Pest Control Applicator training manual | 5

Applicator training manual

Authors Acknowledgments Diane Clark, University of California, Davis This manual was developed using pesticide safety education materials, Extension Janet Fults, Oregon Department of manuals and publications from Texas Agriculture, Pesticides Division Cooperative Extension, the University of Laurie Gordon, Oregon Department of California at Davis; the Universities of Agriculture, Pesticides Division Georgia, Illinois, Kentucky, Tennessee, Missouri and Nebraska; and from Iowa Patrick J. O’Connor-Marer, University of State, North Carolina State, Oklahoma California Davis State and Washington State Universities. Mark Matocha, The Texas A&M University Oregon Department of Agriculture System reproduced portions of this manual with Roland Maynard, Oregon Department of permission of the copyright holders: The Agriculture, Pesticides Division Texas A&M University System and The Regents of the University of California. Douglass E. Stevenson, The Texas A&M Further reproduction of this manual is University System prohibited without written permission of the copyright holder. Contact Oregon Reviewers Department of Agriculture for more information. Rich Affeldt, Oregon State University, Jefferson County Cooperative Extension Service Craig Collins, Collins Agr Consultants Joe DeFrancisco, Oregon State University, North Willamette Research and Extension Center Gale Gingrich, Oregon State University, Cooperative Extension Service (retired) Mark Mellbye, Oregon State University, Linn County Cooperative Extension Service Steve Riley, Oregon Department of Agriculture, Pesticides Division 6 | Demonstration and Research Pest Control Table of Contents | 7

Table of Contents

About this manual 3

Applicator training manual 5

1 Demonstration and research pest control laws and regulations 11

2 Good laboratory practice standards (GLPS) 21

3 Pesticide-organism interactions 25

4 Equipment calibration 29

5 Research and the scientific method 31

6 Planning experiments 33

7 Experimental design 35

8 Data collection 49

Appendix A - Review questions 55

Appendix B - Answers to review questions 61

Appendix C - Glossary 63

Appendix D - Table of random rumbers 65

Appendix E - Latin square design 67

Appendix F - Pesticide contacts 69 8 | Demonstration and Research Pest Control Table of Contents | 9 10 | Demonstration and Research Pest Control 1 Demonstration and research pest control laws and regulations | 11

Demonstration and research pest control laws and regulations Introduction and Health, the U.S. Fish and Wildlife 1 Service, and the U.S. Food and Drug State and federal laws and regulations Administration may also monitor and govern the manufacture, sale, transportation, regulate some activities involving pesticide and use of pesticides. At the national level, use. the United States Environmental Protection Pesticide applicators in all states must Agency (EPA) is the primary pesticide comply with federal requirements. Oregon regulatory agency. In Oregon, the State has enacted additional laws and regulations Department of Agriculture (ODA) assumes that strive to address pesticide use under this role for most pesticide related activities. the special conditions existing here. Oregon The EPA’s authority comes from the Federal requirements are sometimes more restrictive Insecticide, Fungicide, and Rodenticide Act than federal requirements but can never (FIFRA). Originally passed in 1947, this allow uses or activities prohibited at the law has undergone several amendments federal level. and updates, including the 1992 Worker Protection Standard and the Food Quality ODA is the state lead agency in Oregon Protection Act of 1996. Based on FIFRA, for regulating all aspects of pesticide the EPA establishes regulations for pesticide registration, distribution, and use. This registration and labeling, and for pesticide strict oversight begins with the registration residue tolerance levels on or in food or of each pesticide product and continues feed. These regulations also set standards for with the certification and licensing of experimental use of pesticide compounds pesticide applicators, applicator trainees, and for certifying commercial and private dealers, and consultants. Several other state pesticide applicators to use restricted-use agencies in Oregon cooperate with ODA pesticides. Other federal agencies, including in regulating pesticide use. These include the U.S. Department of Agriculture, the the Oregon Departments of Forestry, National Institute for Occupational Safety Transportation, Fish and Wildlife, Human Services, and Environmental Quality. Oregon Occupational Safety and Health Administration is also actively involved in pesticide regulation in Oregon. New laws and regulations can be created to address situations not covered by existing laws and regulations. For example, pesticides and pesticide application equipment are constantly being improved or modified, and these improvements often require people to use pesticides differently. In addition, pest problems and pest management techniques can differ from year to year. This affects how people use pesticides. Law and regulation changes can also address emerging health and environmental problems. For instance, the Worker Protection Standard was a change in FIFRA that strengthened requirements to protect agricultural and forest workers who handle pesticides or otherwise work in pesticide-treated areas. Experimental pesticides applied with an airblast sprayer. Photo courtesy of Laurie Gordon, ODA. If you are involved in demonstration and research uses of pesticides in Oregon, you 12 | Demonstration and Research Pest Control

need to understand and follow general laws and regulations regarding pesticide use. For Note: Certain uses of a pesticide product example: are not considered to be a use in conflict ♦♦ Follow the requirements on the pesticide with the label. Examples are using a pesticide container label or any supplemental product against an unnamed pest (except labeling. for rodenticides and antimicrobial pesticides ♦♦ Ensure worker safety and complying which are pest specific) or using it at a rate with the Worker Protection Standard. lower than the label specifies, provided all ♦♦ Obtain the correct pesticide applicator license to conduct experimental pesticide other use directions and sites are followed. use. Therefore, demonstration trials conducted ♦♦ Protect people, animals and the according to the Oregon-registered label, are environment. not considered experimental and would not ♦♦ Keep pesticide application records. require an EUP. ♦♦ File pesticide use reports, if applicable. ♦♦ Store, transport and dispose of pesticides different methods and systems and can be and pesticide containers properly. used to convey simple recommendations. The results of demonstrations have a more ♦♦ Follow specific laws and regulations limited use than the results of research that cover the types of activities often experiments, because there is no reliable way involved in demonstration and research to measure their limits of confidence. settings, including experimental uses of unregistered products and/or uses that Demonstrations are usually one of the two are not allowed by the label of pesticides types: registered in Oregon. 1. Method demonstrations show how to do something, for example, how to calibrate Demonstration vs. a sprayer. research experiments 2. Results demonstrations show, by example, what happens with Demonstrations and research experiments the practical application of new have different goals and are designed and information, or they show principles or analyzed differently. comparisons that support a practice or recommendation. Demonstrations You should make observations and take Demonstrations of pesticide products are notes throughout the season or term of the usually set up to show how a product or demonstration, particularly on unexpected method works under local conditions or developments. Usually field meetings to compare several products side-by-side. show off the results. An effective result Demonstrations are distinguished from demonstration requires a clear-cut and research trials in that they only involve simple objective and a uniform field site that pesticide uses that are currently allowed by is easy to access. the pesticide label. Examples of topics for demonstrations include the effective use of a piece of application equipment or methods for properly incorporating a herbicide Experimental Pesticide Use into the soil. Rather than requiring careful data collection or analysis, demonstrations Use of any substance or any combination of effectively persuade by being visually substances as a pesticide with the intent convincing. Demonstrations are usually at of gathering data needed to satisfy pesticide one site, short-term, and not replicated. For these reasons, demonstrations tend to be registration requirements of EPA or of easier and less expensive to conduct than ODA shall be considered a pesticide use for research experiments. They can stimulate experimental or research purposes. applicators and growers to think about 1 Demonstration and research pest control laws and regulations | 13

Research experiments minimizing bias and error and maximizing the utility of your results. When you conduct pesticide research experiments your goal is to generate data Research uses of pesticide products that can be used to: might involve experimental or numbered compounds, new formulations, federally ♦♦ Support new pesticide uses or methods, registered products not yet registered in such as new rates, sites, equipment or Oregon, new uses of registered products application frequencies. (e.g., new crops, timing, rates, or methods), and some uses of genetically modified crops. ♦♦ Add new target pest species to the label. ♦♦ Support existing knowledge. Licensing ♦♦ Close gaps in existing information. You must be an Oregon licensed pesticide ♦♦ Develop new information. applicator, trainee or consultant with the Unlike most demonstrations, pesticide Demonstration and Research category research experiments demand careful design to make applications of pesticides used and close management, and you must experimentally or in research situations. conduct data collection and analysis in a Licensing requirements are applicable way that will produce scientifically sound to anyone conducting pesticide research conclusions. including state and federal agency employees who are not required to obtain Most research experiments follow these an EUP. Only a licensed Commercial or fundamental steps: Public Pesticide Applicator may supervise ♦♦ Formulate a hypothesis (a suggested applications made by a licensed pesticide solution or explanation for a specific trainee. problem or question). In 2009, a category of Demonstration ♦♦ Design an experiment to objectively test and Research was incorporated into the your hypothesis. Oregon pesticide licensing program. Anyone conducting experimental pesticide ♦♦ Collect data from the experiment. trials must have this category on their ♦♦ Interpret the data you have collected. pesticide license issued by the department. However, this category is not appropriate ♦♦ Accept, reject, or alter your original for non-pesticide related research, such as hypothesis. plant breeding trials or soil fertility trials, Try to keep your research experiments even if pesticides will be used during the as simple as possible while still satisfying experiment. the needed level of scientific soundness. Different licensing options are available Experimental error and bias are inherent depending upon what types of pesticide in any experiment; therefore, one of your related activities you want to be able to primary goals as a researcher is to reduce do. The following licensing options are these to a minimum. Good experimental available: design and technique go a long way toward Pesticide Consultant license with Demonstration and Research category Maintenance applications are made for This license allows the licensee to provide routine purposes are not covered by the technical advice on any restricted-use Pesticide Consultant license. These include pesticide and make demonstration and any application made in accordance with research applications. A person with this license is prohibited from supervising the pesticide label where the use is not pesticide trainees and may not make routine an experimental factor. An example might “maintenance” pesticide applications. be conducting routine weed control in a Commercial Pesticide Applicator with fungicide trial. Demonstration and Research category

This license option works well for persons who have a pesticide research business or are 14 | Demonstration and Research Pest Control

contracted by others to conduct pesticide ♦♦ The supplier of pesticide product(s) applications for experimental purposes. The applied. licensee can add other specific use categories ♦♦ The trade name and the strength of such to this license to make routine maintenance pesticides applied. applications. The licensee may also supervise licensed trainees only in the categories listed ♦♦ The amount or concentration (pounds or on the license. gallons per acre of active ingredient or concentration per 100 gallons). Public Pesticide Applicator with ♦♦ The specific property, crop(s), or site(s) to Demonstration and Research category which the pesticide was applied. This license option works well for persons ♦♦ The summary information of equipment, who are employed by a public agency device or apparatus used and, if applied (including universities and colleges) to by aircraft, the F.A.A. number. conduct pesticide research, implement ♦♦ Name of applicator(s) and/or trainee(s) pesticide demonstration trials, or are who made the application. contracted by others to conduct pesticide applications for experimental purposes Records must be kept for three years. as a part of their employment with their public agency. The licensee can add other Pesticide Use Reporting specific use categories to this license to expand their application sites to make System (PURS) routine maintenance applications in another category. The licensee may also supervise The Pesticide Use Reporting System licensed trainees only in the categories listed (PURS) is scheduled to be implemented on the license. in 2013. It is anticipated that PURS will require reporting of all pesticide applications Private Pesticide applicator license (including experimental use) through the ODA web site. Details are available on the The consultant, commercial applicator and ODA web site: public applicator licenses are not appropriate for the personal use of restricted-use http://oregon.gov/ODA/PEST pesticides. However, under the Private Pesticide Applicator License you may Crop destruct obtain and use restricted use pesticides for agricultural or forestry production on Crop destruction is one of the most land you or your employer owns or rents. important parts of experimental pesticide This license type is not appropriate for use. You must adhere to these requirements conducting experimental pesticide trials. not only because they are the law, but failing to adhere to them could cause harm to Recordkeeping people, animals, or the environment. Unregistered pesticide products and Records of pesticide applications made for numbered compounds usually lack an experimental or demonstration purposes established EPA residue tolerance for the must be maintained by all licensed pesticide active ingredient and crop combination. applicators, including trainees. Licensed Registered products used experimentally on Pesticide Consultants and licensed crops or in ways not allowed by the label Commercial or Public Pesticide Applicators may exceed or lack existing tolerances. must prepare and maintain the following The EUP holder must destroy the food or information: feed item unless: ♦♦ The firm or individual for whom the 1. A residue tolerance has been established pesticide application was made. by EPA for the pesticide-crop ♦♦ The location of the land or property combination, rate, and use pattern you where application was made. are testing, or ♦ 2. ♦ The date and approximate time of The pesticide you are testing has been application. exempted from the requirement of a residue tolerance, or 1 Demonstration and research pest control laws and regulations | 15

chain without complying with established tolerances. Documentation of crop destruct becomes very important when there could be a trace-back of contaminated food/feed items to the trials you are responsible for. Documentation must include the date and the crop destruct method. Photos of the destruction process are also recommended. Grazing restriction If a crop has been treated with a pesticide that does not have a tolerance established for forage or for meat and milk, the treated site must not be used for grazing of animals for a minimum of 365 days from the date of the last application. The permit holder must ensure that the grower/cooperator is Grazing of test plots is prohibited for 365 days from the last application. Photo source: ODA archives informed of this restriction. If animals are allowed to graze on land that 3. The pesticide you are testing has a time- has been treated with a pesticide that does limited tolerance established by EPA not have a tolerance for such, the animals that is in effect. may be harmed or may have pesticide residues in the animal’s meat or milk. In Crop destruction means “to render unusable this situation, the animals, meat, milk or for food or feed, or to use for research other contaminated commodity may not be purposes only.” The crop destruction rule consumed or marketed for any purpose. applies to all treatments for crops, including dormant, fallow, and pre-plant treatments. No portion of a crop to which a pesticide Treated crops grown for product having no established pesticide seed residue tolerance for the crop has been applied, shall be used or distributed for food If you grow a specialty seed crop (other or feed. This restriction pertains to, but is than grass grown-for-seed) for a research not limited to, green chop, hay, pellets, meal, experiment or demonstration, you need to whole seed, cracked seed, straw, roots, bulbs, know your responsibilities. Know whether foliage or seed screenings. This restriction pesticides used on specialty seed crops have also includes grazing the crop, stubble, or established pesticide reside tolerances. If re-growth for 365 days. If you submit a a tolerance is lacking or the labeled rate justification for harvest or use, you must is exceeded, you need to inform your seed include information about the pesticide conditioner which pesticides were applied to product’s applicable residue tolerances, your crop. It is your responsibility to ensure tolerance exemption, or the federally no portion of a treated seed crop is used or registered label that allows such use. distributed for human food or animal feed. When certain EUPs are authorized by All seed harvested from plants that have ODA for an experimental pesticide use been treated experimentally must be labeled on a food or feed crop, ODA issues a with the following statement: Notice of Detainment (embargo notice) prohibiting the use of any part of the crop “This seed was produced using one or treated experimentally with pesticides. The more products for which the United States EUP holder must explain the implications Environmental Protection Agency has not of a Notice of Detainment to the grower established pesticide residue tolerances. This or person in control of the site. In other seed, in whole, as sprouts, or in any form, cases ODA may not issue a Notice of may not be used for human consumption Detainment, but the EUP holder is still or animal feed. Failure to comply with this required to ensure that the crop is destroyed condition may violate the requirements of when appropriate. Food or feed plant parts the Federal Food and Drug Administration, are never allowed to enter the food or feed the Oregon Department of Agriculture, and other regulatory agencies.” 16 | Demonstration and Research Pest Control Residue testing Acreage Limitations If your trial uses a product that is a federally Under an Oregon EUP, the use of a pesticide registered pesticide for your crop and you experimentally is limited to a cumulative total have used the product according to its label, it is presumed to meet the required residue of 10 acres per chemical (active ingredient) tolerance without further testing. If you per year. Testing on areas larger than this treat crops at higher than labeled rates, by a requires that a federal EUP be obtained prior different application method, or use shorter pre-harvest intervals (PHI), generally you to requesting an Oregon EUP. On some rare must destroy the crop. However, in some occasions, ODA may request EPA to waive instances ODA may allow the crop to be the federal EUP requirement to address a analyzed to confirm that it is within the established tolerance. If confirmed, ODA specific state need to conduct experimental would then provide a written release of pesticide use on acreage over 10 acres. Upon embargo on the crop before it is harvested EPA approval, an Oregon Site-Specific EUP for food or feed purposes. would be issued. Federal and state Oregon. ODA will determine if the trial will experimental use permits be permitted in Oregon and if so, a Site- (EUPs) Specific Oregon EUP will be issued. Federal EUP regulations require pesticide Under Oregon law, if you are planning to products shipped or used under federal EUP use certain pesticide products for research to be labeled with directions and conditions purposes you may be required to obtain for use. In most cases, this labeling will an authorization to conduct such research include the following: called an Experimental Use Permit (EUP). In some cases involving pesticide registrants ♦♦ A prominent statement “For conducting pesticide development research Experimental Use Only.” for EPA registration purposes, a federal ♦♦ The federal EUP number. permit for experimental use of pesticide products may also be required. The following ♦♦ The name, brand, or trademark. sections outline the requirements for ♦♦ The name and address of the permit Oregon’s and the EPA’s Experimental Use holder, producer, or registrant. Permits (EUPs). ♦♦ The net contents. Federal EUP ♦♦ An ingredient statement. ♦ Federal Experiment Use Permits ♦ Any appropriate limitations on entry of people into treated areas. The US EPA administers the Federal EUP ♦♦ The establishment registration number Program. This permit is sought by entities except in cases where application of the that want to conduct experimental use on pesticide is made solely by the producer. more than 10 acres or more than 1 surface ♦ acre of water in Oregon. Experiments with ♦ The directions for use. pheromones used at rates less than 150 ♦♦ In addition to these items, when a federal grams ai/acre/year require a Federal EUP EUP is used under federal conditional when the site exceeds 250 acres. Visit the registration, the labeling must include EPA web site for information on applying the following statement: “Not for sale for a Federal EUP: http://www.epa.gov to any person other than a participant or cooperator of the EPA approved Using a Federal Experimental Use Permit in Experimental Use Program.” Oregon There are situations in which it is not legally File an application with ODA at least 30 necessary for a pesticide manufacturer to days prior to initiation of any trials involving obtain a federal EUP, however, an Oregon a Federal Experimental Use Permit in EUP would still be required and trials are limited in size and scope. In these cases, 1 Demonstration and research pest control laws and regulations | 17

the EPA presumes that the nature and ♦♦ Appropriate pesticide applicator, size of the experiment will not involve consultant, or trainee licensing. unreasonable adverse effects. Examples ♦♦ Pesticide application recordkeeping. and guidelines that help determine when a federal EUP might not be necessary are ♦♦ Pesticide reporting, if applicable. outlined in the Code of Federal Regulations ♦♦ Adherence to crop destruct provisions if (40 CFR Section 172.3). no tolerance has been established for the Oregon Experimental Use Permits pesticide applied to a specific crop. ♦♦ Adherence to grazing restrictions of 365 (EUPs) days if no tolerance has been established for the pesticide applied to a specific crop Oregon has specific regulations regarding or site (pasture). pesticide experiments, including unregistered uses of pesticide products. Two Oregon EUP Types Before you conduct experiments with pesticides (including pheromones), you are Two distinct EUPs are available to address required to obtain an EUP from ODA, experimental or research uses of pesticides unless you are specifically exempted (see in Oregon. “Oregon EUP Exemptions). Oregon EUPs are required for all sizes of Collective Experimental Use Permit experiments, including those uses covered Features: by a federal EUP. Always understand your responsibilities when conducting trials ♦♦ Allows a researcher who conducts involving pesticides. An EUP is not an small plot trials to apply with ODA to exemption from pesticide registration. The cover all trials conducted that year. This purpose of the EUP program is to: process supports small plot research with minimum regulatory oversight. ♦♦ Allow for research that contributes to the product registration process. ♦♦ Covers experimental pesticide use that does not exceed a total of one acre per ♦♦ Ensure that research-related crops or test pesticide and is conducted on agricultural animals bearing illegal residues do not or forest sites. enter the food chain. ♦♦ Does not allow for aquatic pesticide ♦♦ Protect humans, animals and the trials or experiments involving animals or environment. livestock. Oregon EUP Exemptions ♦♦ Expires December 31st of the year in which issued. If you are currently employed by a federal or state agency and will be conducting Requirements: research under their established policies for ♦♦ Provide a summary report no later pesticide use, obtaining an Oregon EUP is than 30 days after the expiration of the not required. However, always check first permit. with your agency to find out about their ♦ particular pesticide use and experimentation ♦ Keep records of pesticide application(s). policies. Some agencies may require you to ♦♦ Document crop destruction, if applicable. obtain an EUP even if the state regulations ♦ do not. ♦ File pesticide use reports, if applicable. Pesticide research conducted entirely within Site-Specific Experimental Use Permit a greenhouse is also exempt from the requirement to obtain an Oregon EUP. Features: Exemption from the requirement of ♦♦ Is for experimental pesticide use on obtaining an Oregon EUP does not exempt specific sites other than agriculture or a person from their responsibilities under forest or any site that exceeds one acre in state and federal law to comply with other total size. aspects of experimental use of pesticides. ♦ ♦ Approvable sites include, but are These responsibilities include: not limited to, aquatic, residential, recreational and structural sites, areas 18 | Demonstration and Research Pest Control

with public access, commodity storage ♦♦ Identification of each pesticide to be facilities, animals/livestock, and any area used, including: 1) the names of the exceeding a total of one acre. active ingredients, 2) the product names (if any), and 3) the EPA Registration ♦♦ Expires no more than 12 months after number (if any). issuance by the department. ♦♦ The number of the current ODA Requirements: pesticide-related license of the person(s) ♦♦ Research must provide a summary report making the pesticide applications and no later than 30 days after the expiration the means of locating the person in case of the permit. of an emergency. ♦♦ Keep records of pesticide applications. ♦♦ The purpose of the experiment or research, including a list of the intended ♦♦ Document crop destruction, if applicable. target pest(s). ♦♦ File pesticide use reports, if applicable. ♦♦ The approximate date(s) of pesticide use. Applying for an Oregon EUP ♦♦ Specific description and location of each site where pesticide use may occur, ODA’s EUP application forms are available including the size (for example; acres or by calling 503-986-4635 or via ODA’s web square feet) of each site. site at http://oregon.gov/oda/pest. ♦♦ Disposition of any food or feed item Submit the completed “Collective” or from the crop or site on which the “Site Specific” EUP application form to pesticide will be used (if no tolerance has ODA for approval at least 30 days prior to been established). intended use. ODA may require additional ♦ information in order to adequately assess ♦ Application rate(s) of the pesticide and the potential adverse effects on workers, the number of applications. public, or the environment. ♦♦ Method of application. Each Collective Experimental Permit ♦♦ Timing and duration of the proposed application must include: experiment or research. ♦♦ Name, address, and telephone numbers ♦♦ Total amount of pesticide to be used, of the permit applicant. diluent and dilution rate. ♦♦ Name, address, telephone numbers and ♦♦ Copy of any federal EUP issued by EPA. pesticide applicator or consultant license ♦ numbers of the person(s) responsible ♦ Copy of the labeling that will accompany for carrying out the provisions of the the pesticide in the field. experimental use permit at each specific ♦♦ Documented support of the pesticide site and the means of locating the person registrant. in case of an emergency. ♦♦ Adverse properties of active ingredients. ♦♦ A signed statement that all pesticide use ♦ will comply with all of the provisions of ♦ Crop/site. the collective experimental use permit ♦♦ Pre-harvest interval (PHI). and of this section ♦♦ Restricted-entry interval (REI). Each Site-Specific Experimental Use ♦ Permit application must include: ♦ Significant bordering landmarks. ♦♦ Name, address, and telephone numbers of the permit applicant. EUP notification and other ♦♦ Name, address, telephone numbers and requirements pesticide applicator or consultant license numbers of the person(s) responsible for carrying out the provisions of the 72-hour advance notice experimental use permit at each specific For Site-Specific EUPs Only: You must site and the means of locating the person notify ODA at least 72 hours before the first in case of an emergency. pesticide application is made. ODA prefers to be notified by e-mail. Fax or voicemail 1 Demonstration and research pest control laws and regulations | 19

regulations that are relevant to your Adverse effect reporting trial. When you work with pesticides in The permit holder or person that conducted experimental trials, you are responsible for ensuring the legal and safe use of the the pesticide use must immediately report materials you use. to the Department any adverse, environmental, human or animal effects resulting from Storage and disposal pesticides used for experimental or research When you use registered pesticides in your purposes. trials, follow the requirements on the label about how to store and dispose of unused pesticides and empty pesticide containers. If notification is also accepted. When your trial involves an unregistered material, providing notification, please give the EUP follow the storage and disposal guidelines number and location of the application on the product’s Material Safety Data Sheet site intended to be treated and the date/ or technical bulletin. approximate time when application is to be made. Special conditions, Summary report regulations, and The Oregon EUP process requires you to enforcement file a summary report within 30 days of the expiration of any EUP. This report must Each EUP approved by ODA may contain be filed by persons that have either type of specific conditions under which the research Oregon EUP (Collective or Site-Specific). is to be conducted. ODA will include those Each summary report must include, at a conditions on the approved EUP. minimum, the identification number of ODA has the authority to terminate, the experimental use permit, the record amend, or refuse to issue an EUP under information required to be maintained, any certain circumstances, including but not adverse effects to humans, animals or the limited to: environment and documentation of crop ♦ destruct, if applicable. ♦ If the research involves a hazard to pesticide handlers, field workers, public Use precautions health, animals or the environment. ♦♦ If the research is used for purposes If your EUP trial uses a registered pesticide, unrelated to pesticide registration data follow the use precautions and protective development. clothing requirements on the label. ♦ However, if your trial involves the use of ♦ If the applicant has failed to provide the experimental compounds whose toxicity has required summary report for a previously not been fully evaluated, you should use at issued EUP. least the following protective clothing: ODA Pesticides Division staff routinely ♦♦ Protective eyewear. monitor EUP trials that are conducted to ensure compliance with the EUP ♦ ♦ Long-sleeved shirt. requirements and conditions of approval. ♦♦ Long pants. If any adverse effects of the trial or any noncompliance with the conditions of the ♦ ♦ Chemically-impervious gloves. permit or of the Oregon Revised Statutes ♦♦ Chemically-impervious boots. Chapter 634 (ORS 634) are determined, you risk enforcement action including fines ♦ ♦ Other protective clothing indicated on and/or pesticide license action. a technical bulletin or Material Safety Data Sheet. In addition to any other liability or penalty provided by law, any failure by any person to ♦ ♦ Certain pesticides may have additional comply with the provisions of this section, regulations regarding worker exposure, as determined by the Department, may restricted-entry intervals, and other be used as a basis of one or more of the restrictions or limitations of use. following actions: You must be aware of and follow all 20 | Demonstration and Research Pest Control

♦♦ To revoke or suspend or refuse to issue an experimental use permit. ♦♦ To revoke or suspend or refuse to issue any license of a permit holder or of a person that conducted a pesticide use for experimental or research purposes. ♦♦ To impose a civil penalty. 2 good laboratory practice standards (GLPS) | 21

Good laboratory practice standards (GLPS) The most important regulation for pesticide GLPS definitions 2 researchers submitting data to EPA in support of FIFRA and Toxic Substances 40 CFR 160.3 defines the following Control Act (TSCA) actions is 40 CFR important terms for pesticide studies. 160, EPA’s Good Laboratory Practice Standards or GLPS compliance monitoring Study program. It applies to all researchers generating test data that will be submitted Study means any experiment at one or more to EPA under FIFRA and TSCA. This test sites, in which a test substance is studied may be the most important regulation that in a test system under laboratory conditions applies to people doing pesticide field tests or in the environment to determine in support of FIFRA section 3 registrations, or help predict its effects, metabolism, section 24c registrations or section 18 uses. product performance (efficacy studies EPA’s GLPS compliance monitoring only as required by 40 CFR 158.640), program ensures the quality and integrity environmental and chemical fate, persistence of test data submitted to the agency under and residue, or other characteristics in FIFRA and TSCA. All researchers who humans, other living organisms, or media. conduct a study that will be submitted to EPA must comply with GLPS regulations Test substance located at 40 CFR part 160 for data under FIFRA and 40 CFR part 792 for data under Test substance means a substance or mixture TSCA. administered or added to a test system in a study. Failure to comply with GLPS (40 CFR 160) is an actionable offense that can result Test system in cancellation, suspension or modification of the registration, research or marketing Test system means any animal, plant, permit; the imposition of civil penalties; or microorganism, chemical or physical matrix, criminal prosecution under 18 U.S.C. 2 or including but not limited to soil or water, or 1001, or FIFRA section 14. subparts thereof, to which the test, control, or reference substance is administered or added for study. Testing facility Testing facility means a person who actually conducts a study, i.e., actually uses the test substance in a test system. “Testing facility” encompasses only those operational units that are being or have been used to conduct Researchers gather data and take notes on studies. a field experiment. Experiments conducted under GLPS have specific documentation requirements. Photo courtesy of Joe Control substance DeFrancesco, OSU NWREC. Control substance means any chemical substance or mixture, or any other material other than a test substance, that is administered to the test system in the course of a study for the purpose of establishing a basis for comparison with the test substance. 22 | Demonstration and Research Pest Control

Carrier 1. Descriptive title and statement of purpose. Carrier means any material, including but 2. not limited to feed, water, soil and nutrient Identification of the test, control, and media, with which the test substance is reference substance by name, chemical combined for administration to a test abstracts service (CAS) number or code system. number. 3. The name and address of the sponsor Raw data and the name and address of the testing facility at which the study is being Raw data means any laboratory worksheets, conducted. records, memoranda, notes, or exact copies 4. The proposed experimental start and thereof that are the result of original termination dates. observations and activities of a study and are necessary for the reconstruction and 5. Justification for selection of the test evaluation of the report of that study. system. 6. In animal studies, where applicable, the Specimens number, body weight range, sex, source of supply, species, strain, substrain, and Specimens means any material derived from age of the test system. a test system for examination or analysis. 7. The procedure for identification of the Sponsor test system. 8. A description of the experimental Sponsor means: 1) A person who initiates design, including methods for the and supports, by provision of financial or control of bias. other resources, a study; 2) A person who 9. submits a study to the EPA in support of Where applicable, a description and/ an application for a research or marketing or identification of the diet used in the permit; or 3) A testing facility, if it both study as well as solvents, emulsifiers and/ initiates and actually conducts the study. or other materials used to solubilize or suspend the test, control, or reference Study director substances before mixing with the carrier. The description shall include Study director means the individual specifications for acceptable levels responsible for the overall conduct of a of contaminants that are reasonably study. expected to be present in the dietary materials and are known to be capable of For all definitions pertinent to GLPS, interfering with the purpose or conduct consult 40 CFR 160.3, in which terms are of the study if present at levels greater listed in alphabetical order. than established by the specifications. Study protocol 10. The route of administration and the reason for its choice. requirements 11. Each dosage level, expressed in milligrams per kilogram of body or Fifteen specific items of information are test system weight or other appropriate required for every study to which GLPS units, of the test, control, or reference applies. The regulation states that “each study shall have an approved written protocol that clearly indicates the objectives and all methods for the conduct of the Changes or Revisions study.” The protocol for a study must contain but is not necessarily limited to All changes in or revisions of an approved these items of information. Note. The GLP protocol and the reasons for them records requirements under GLPS differ shall be documented, signed by the study from the Oregon pesticide recordkeeping requirements for consultants and director, dated, and maintained with the commercial and public applicators. Both sets protocol. of records should be maintained separately. 2 good laboratory practice standards (GLPS) | 23

substance to be administered and the All persons conducting tests of pesticides method and frequency of administration. must comply with protocols and regulations governing data handling, storage and 12. The type and frequency of tests, analyses, retrieval; records of equipment maintenance and measurements to be made. and calibration; and the transfer, proper 13. The records to be maintained. placement, and identification of test systems. They must have immediately available 14. The date of approval of the protocol by manuals, protocols and standard operating the sponsor and the dated signature of procedures relative to the laboratory or field the study director. procedures being used in the tests (40 CFR 15. A statement of the proposed statistical 160.81). method to be used. All raw data, documentation, records, Section 160.130 of 40 CFR requires that protocols, specimens, and final reports the study shall be conducted in accordance generated as a result of a study shall be with the protocol. The test systems shall be retained. Correspondence and other monitored in conformity with the protocol. documents relating to interpretation Specimens shall be identified by test and evaluation of data, other than those system, study, nature, and date of collection. documents contained in the final report, This information shall be located on the also shall be retained. There shall be archives specimen container or shall accompany the for orderly storage and expedient retrieval specimen in a manner that precludes error of all raw data, documentation, protocols, in the recording and storage of data. specimens, and interim and final reports (40 CFR 190). All records relating to a test shall Section 160.130(e) of 40 CFR requires that be retained for a period of at least 5 years all data generated during the conduct of a following the date on which the results of study, except those that are generated by the study are submitted (40 CFR 195). automated data collection systems, shall be recorded directly, promptly, and legibly in For detailed information on compliance ink. All data entries shall be dated on the and monitoring of GLPS, consult 40 day of entry and signed or initialed by the CFR 160, EPA Manual 2185 Good person entering the data. Any change in Automated Laboratory Practices, and EPA entries shall be made so as not to obscure Manual 723-B-93-001 Good Laboratory the original entry, shall indicate the reason Practice Standards Inspection Manual, or for such change, and shall be dated and the American Chemical Society’s Good signed or identified at the time of the Laboratory Practice Standards: Applications change. for Field and Laboratory Studies. 24 | Demonstration and Research Pest Control 3 Pesticide-organism interactions | 25

Pesticide-organism interactions

Federal regulations (40 CFR 171.4(c) population to a pesticide. Many insect 3 (10)) require that persons conducting and mite species have become resistant to demonstration and research work pesticides worldwide. In addition, at least with pesticides should demonstrate an 200 species of fungi, more than 200 species understanding of pesticide-organism of weeds, and several species of nematodes interactions and their importance in IPM and rodents also are resistant to one or more programs. pesticides. Both the beneficial and harmful effects Resistance often develops in pest of pesticides are determined by pesticide- populations that have been treated organism interactions. To be effective, a frequently with pesticides that have a pesticide must 1) penetrate the organism, 2) common mode of action. The development move or be transported to the site of action, of resistance may sometimes be averted and 3) disrupt or alter a vital function. The or delayed by reducing the number of manner in which the pesticide affects the treatments and/or alternating the use of vital function is called its mode of action. pesticides with different modes of action. Penetration, transport and mode of action involve pesticide-organism interactions. Penetration Pesticide-organism interactions also are involved in the metabolism, accumulation The speed and extent of penetration depends and elimination of pesticides by the on the permeability of the organism to organism, and in biodegradation and the specific pesticide. Permeability differs biological magnification. significantly among plants and insects and even among different tissues of the same organism. Among animals, tissues Resistance of the respiratory and digestive system are usually much more permeable than the Pesticide selectivity and the development skin. In plants, new, succulent growth is of pesticide resistance are often caused more permeable than hardened growth and by differences in pesticide-organism bark. The ability of a pesticide to penetrate interactions. Selectivity is the ability of an organism depends on its chemical a pesticide to affect one organism and nature and the formulation. Penetration not another. Resistance is the heritable can sometimes be increased by either reduction in the sensitivity of a pest incorporating adjuvant compounds directly into a formulated pesticide product (by the manufacturer) or adding an adjuvant product to the diluted pesticide mixture in the tank. Transport The ease with which a pesticide moves from the place where it entered an organism to its site of action depends on the mobility of the pesticide molecules and the efficiency of the transporting mechanism of the plant or animal. For example, systemic herbicides move throughout the plant, while other herbicides are not mobile and affect only the Figure 1. Pesticide resistance increased rapidly with the expansion of pesticide area they contact directly. technology during the last half of the 20th century, but accelerated exponentially after 1975, when the number of registered pesticides began to decline. 26 | Demonstration and Research Pest Control Mode of action A pesticide performs its main function only after it reaches action sites within an organism. For example, organophosphate insecticides target nerve cells and the herbicide atrazine affects photosynthesis in the chloroplasts of plant cells. Pesticides kill or otherwise alter an organism by disrupting some vital physiological function. This is known as the pesticide’s mode of action. Organophosphate insecticides (e.g., methyl parathion, malathion, phorate) inhibit the breakdown of acetylcholine by cholinesterase, an enzyme that helps regulate the nervous system. This causes muscles and glands to become overactive because nerve cells are Figure 3. Sequestration, storage and accumulation of toxins in insect pests is a method of deactivating environmental and dietary poisons that also confers a over-stimulated. Some herbicides act as passive defense against predators on insects that can use it. plant growth regulators, speeding up or slowing down cell growth and reproduction; alter the compound, leaving the toxic 2,4-D other herbicides may target vital plant metabolite. functions or specific enzymes. One type of herbicide, the ALS inhibitors, blocks Given enough time, an organism may be the synthesis of an enzyme that is critical able to metabolize certain pesticides to less to the production of several amino acids. toxic metabolites. Survival may depend on Fungicides may inhibit spore germination whether or not the organism can metabolize and fungal growth. the pesticide into less toxic metabolites before the toxic activity is complete or Metabolism irreversible. Metabolism is the process by which a Accumulation, elimination pesticide or other chemical is changed into one or more different chemicals within a and storage living organism. The metabolic product, or Pesticides and their metabolites may metabolite, may be either more toxic or less be stored or accumulated within an toxic than the original pesticide ingredient. organism, or they may be eliminated as Aldicarb, the active ingredient in Temik®, waste. If the level of exposure to most has metabolites that are as toxic as or accumulated pesticides remains constant, an slightly less toxic than aldicarb itself. equilibrium between storage, metabolism Some pesticides are effective only after and elimination is reached, and the concentration of the pesticide and its metabolites remains constant within an organism. If the level of exposure is changed, the concentration within an organism correspondingly increases or decreases. Because pesticide residues may accumulate within organisms, producers must take Figure 2. Metabolic activation of 2,4-DB by beta- special precautions during harvest or oxidation to 2,4-D. slaughter. Observing specified intervals between pesticide application and grazing, harvest or slaughter ensures that the they have been metabolized to a lethal products will be safe for consumption. compound. For example, 2,4-DB is changed rapidly to 2,4-D by broadleaf plants (other than legumes). Actually 2,4-DB is relatively harmless to the plant in itself. But enzymes of susceptible broadleaf plants 3 Pesticide-organism interactions | 27 Biodegradation Biological magnification Pesticides in the environment can be Biological magnification is the tendency for affected by certain pesticides to become progressively more concentrated in each type of ♦♦ Soil microorganisms. organism as they move up the food chain. ♦♦ Soil organic matter. An example of biological magnification is when birds of prey become sick after ♦♦ Soil pH. feeding on animals poisoned by pesticides. ♦♦ Soil texture. Perhaps the most familiar example is the thinning of the eggshells of birds exposed ♦♦ Soil moisture. to certain organochlorine insecticides ♦♦ Temperature. such as DDT. This eggshell thinning may result from a chain of events that began ♦♦ Humidity. when invertebrates that consumed plants ♦♦ Ultraviolet light (affects microbial containing DDT residues were, in turn, populations). eaten by rodents, reptiles, amphibians, fish and insectivores, with the residues These factors can affect not only the efficacy becoming more concentrated in each of a pesticide but also the manner and rate species. These intermediate predators in the of its biodegradation—the decomposition food chain were eaten by the top predators, of pesticide residues in the environment by which then received yet higher insecticide bacteria and other microorganisms. concentrations. It is important to be aware of such pesticide-organism interactions when working with certain pesticides. Pesticide interactions Crops may receive several pesticide treatments during a season. Some agricultural chemicals degrade rapidly and do not interact with other chemicals, but some do persist and interact with chemicals that are already present or that are applied later. Most pesticide interactions involve herbicides, which can interact with other herbicides and with non-herbicidal chemicals. These interactions may not affect a herbicide, or they may make it more or less toxic than normal. Synergism occurs when the plant response is greater than expected (more than an additive effect). Antagonism occurs when the plant response is less than expected (less than an additive effect). Herbicide synergists are non-herbicides that are not phytotoxic themselves but are used to increase the phytotoxicity of a herbicide by increasing the amount a plant takes up, preventing the herbicide’s deactivation, or affecting some more complex process. A synergist can be an adjuvant such as a crop oil or surfactant.

Figure 4. The classic example of biological magnification of a pesticide occurred A herbicide antidote is a non-herbicide that in ospreys, the bald eagle, brown pelicans and other fish-eating birds, when decreases a herbicide’s phytotoxicity when DDT accumulated at more than 10 million times its original concentration at mixed with it. Antidotes are used to protect the top of food chains. crops from herbicide injury. An example is the use of a carbon band placed over the 28 | Demonstration and Research Pest Control

seed row to protect the new seedling from the application of diuron herbicide. Research has shown that phytotoxic interactions between major pesticide groups are infrequent, but not rare. Using insecticides with herbicides can increase or decrease the herbicidal activity. Most herbicide-insecticide mixtures increase the injury to the crop. Herbicidal interaction with fungicides is generally antagonistic. 4 equipment calibration | 29

Equipment calibration

The correct calibration of equipment and the measure in milligrams, grams or ounces. 4 accurate measuring and mixing of pesticides Conversion tables are provided in Appendix are extremely important in demonstration A. and research pest control work. In small Because demonstration and research plots the hazards of application may be plots are often relatively small, hand-held reduced and the chances of non-target equipment is usually used for pesticide pollution minimized, but the probability application. The equipment must be of applying a pesticide at the wrong rate calibrated to apply pesticides precisely so is generally greater than in large areas. that research results will be accurate. Before Small errors in measuring the experimental calibration, check all nozzles for uniform material, for example, may cause over- or output. Replace a nozzle if the amount it under-dosing of the treatment plot. An delivers varies more than 5 percent from extra 2 fluid ounces of herbicide added to the average output of all the nozzles on the a 100-gallon tank of water for general field boom. application may not be significant. However, an extra 2 ounces added to 2 quarts of Small sprayers can be calibrated by using the water in small plot research may produce following method: inaccurate or unexpected results. 1. Based on the manufacturer’s broadcast Small-plot experiments often demand that recommendations on gallons per the researcher work with measurements of acre (gpa) application rates, and on grams, milliliters or ounces rather than pints nozzle type for the specific situation, or pounds. In demonstration and research, select a suitable spray tip size from it is not acceptable to use rough estimates or catalogs. Note information on various round off measurements. combinations of pressure, nozzle spacing, ground speed, spray angle, and Liquid measurements should be made with the gallons per minute (gpm) or gpa graduated cylinders or pipettes. Automatic delivered for the nozzle tip you choose. dispensers should be used with pipettes to avoid getting the pesticide into the 2. Select the ground speed based on researcher’s mouth. Dry materials should be what you consider to be a comfortable measured on properly adjusted scales that operating speed. A common speed is 3 miles per hour (mph) for plot work with hand-held sprayers. 3. At this point, the nozzle type, tip size, angle, spacing, pressure, and sprayer speed that closely deliver the recommended gpa rate have been established. Now determine the actual spray delivery in gpm or gpa output by collecting the output from a nozzle over a timed period and by using one of the following formulas. For the greatest accuracy, use a graduated cylinder to collect the sprayer output. gpa x mph x W gpm = 5,940 Experimental sprayers, such as this one, must be calibrated prior to being used in 5,940 x gpm (per nozzle) an experimental setting. Photo courtesy of Mike J. Weaver, Virginia Tech, http:// gpa = pesticidepics.org. mph x W 30 | Demonstration and Research Pest Control

where W = spacing in inches between nozzles on a boom and 5,940 is a constant used in sprayer calibration calculations. If using a single nozzle, W is the spray band width in inches. milliliters per minute ÷ 3,785 ml/gal = gpm Note: Output can be collected for less than a minute. If 180 ml is collected in 30 seconds, what is the output in ml/min? In gpm? 180 ml / 30 sec = 6 ml/sec x 60 sec/min = 360 ml/min 360 ml/min ÷ 3,785 ml/gal = 0.095 gpm Using a consistent walking speed spraying technique makes the calibration of 4. Sample calculations using gpm and gpa backpack sprayers possible. Photo courtesy of Mike J. Weaver, Virginia Tech, formulas http://pesticidepics.org. If your goal is to apply 30 gpa using a CO2 backpack sprayer with a 4-nozzle boom, with 20-inch nozzle spacing, while walking at 3 mph, how many milliliters should be collected in 30 seconds?

30 gpa x 3 mph x 20 gpm = in = 0.3 gpm 5,940

0.3 gpm x 3,785 ml/gal = 1,147 ml/min 1,147 ml/min ÷ 2 = 573.5 ml/30 sec If you collect 0.25 gpm from one nozzle using the same variables as above, what is the sprayer’s gpa rate?

5,940 x 0.25 24.75 gpa gpa = gpm = delivered 3 mph x 20 in 5. Final adjustments in pressure or variation in ground speed can fine tune the output to the desired amount. For major changes in spray output, replace nozzle tips. 5 Research and the scientific method | 31

Research and the scientific method

Research involves the use of the scientific It is important to realize, however, 5 method to discover facts or principles. All that no answer is absolute and that all scientific research is conducted by using the generalizations drawn from an experiment following sequence of steps: should be made with care. ♦♦ Making observations from which to develop a hypothesis. Inference ♦♦ Formulating a hypothesis or a predicted A process called “inference” is used to outcome. test a hypothesis. Inference is the process ♦♦ Designing an experiment to test the of drawing a conclusion based solely on hypothesis objectively. what you already know. It means reading all the clues and making your best guess. ♦♦ Carefully conducting the experiment. These clues come from the analysis ♦♦ Collecting data. of experimental data. When drawing conclusions from an experiment, you will ♦♦ Analyzing and interpreting data. have to use both deductive and inductive ♦♦ Accepting, rejecting or altering the reasoning. original hypothesis based on data Deductive reasoning is inference in which analysis. the conclusion is no more general than the ♦♦ Drawing conclusions (inferences) about experimental data. In essence, it is the failure results. to reject or the rejection of the hypothesis based upon the experimental data. Inductive A well-designed experiment should reasoning is inference in which the premises be simple and precise and contain no of an argument support the conclusion but systematic error (e.g., the plots receiving one do not prove the point. Induction is the treatment should not differ systematically process of drawing broader conclusions from the plots receiving another treatment). from an experiment based upon the failure The researcher should follow the scientific to reject or the rejection of the hypothesis. method meticulously when designing an experiment. Statistical inference involves using data drawn from a portion of an unknown A hypothesis can’t truly be proved because population or process to draw a conclusion you never know if there isn’t one more about a property of that population or experiment that will prove it wrong. For a process. The most common type of inference simple example, say that your hypothesis is involves getting an approximation of a that all the beans in a cloth bag are white. property of interest by collecting a set of You pull out a bean and it is white. Have measurements from a restricted portion of you proved your hypothesis? No, all you the unknown population or process. This have done is not disproved it. If you pull set of measurements is sometimes called out a red bean, you know your hypothesis is a “sample” or sometimes a “representative wrong. If you pull out all the beans you can sample.” prove your hypothesis. However, you can’t do all the possible experiments on a topic; Statisticians have developed formal rules hence the scientific method. for inference from both observational and empirical data. The rules are based on the There is a world of difference between likelihood (probability) that substantially failing to disprove and proving a hypothesis. similar data will result from additional Make sure you understand this distinction. observations made in the same way. In It is the foundation of the scientific method. statistical inference, experimenters use All research relies on it and it is important deduction to accept or reject hypotheses in complying with GLPS requirements. based upon experimental data and induction 32 | Demonstration and Research Pest Control

to generalize from observed information It is important to distinguish producers’ and after the analysis of sample data from researchers’ respective responsibilities. It is experiments. not fair to expect extra work from producers or meticulous station-type data collection On-farm research during busy seasons. The researchers must adapt the experiment to what they judge When planning an experiment, the each producer can conveniently do. researcher has the option of conducting it Decisions must be made on experimental on a research farm or cooperating with a designs, such as the number of treatments private farm. The benefits and drawbacks of and replications or the size of the plots, to each should be considered. accommodate the constraints of a private A research station might be a better farming operation. Producers tend to place to screen riskier alternatives (such favor private farm trials that use standard as unregistered pesticides), to conduct machinery and require little extra time to experiments that may make a field look implement and maintain. bad, or to do experiments that could leave One private farm method that has become producers with a lingering problem, such as acceptable to many producers uses long, weeds. Some pests are best studied under narrow, strip plots. The plots are arranged very controlled conditions on a research in randomized blocks to accommodate farm because of highly variable populations. large machinery. All strips are managed The researcher may decide to conduct identically throughout the growing season an experiment on a private farm if the except for the treatments being tested. farm location offers physical conditions Example that are not found at a research station (e.g., a particular soil type, climate or pest If a trial is testing different methods of infestation). Also, some pest problems may weed control, all strips should receive the be very difficult to create artificially on a same primary tillage, seedbed preparation, research farm (e.g. established perennial fertilizer application, insect control, and weeds). In these cases, private farm studies so on. The only difference in management are more appropriate. over the entire test area would be the weed Other factors may influence a decision to control treatment used on individual strip conduct an experiment at a private farm plots. rather than a research station. Often private Once a cooperator has been selected, efforts farm research is more credible and accessible should be made to compensate him or to producers, particularly if it is done her fairly for time and effort in a manner with large, machine-harvested plots. The appropriate to each individual situation. researcher should keep in mind that the two sites (private farm vs. research station) may require different degrees of intensiveness in monitoring the experimental plots.

The principles of the scientific method should be applied in every experimetal pesticide trial. Photo courtesy of Laurie Gordon., ODA 6 Planning experiments | 33

Planning experiments

In research there are two broad types of All effective experiments involve these basic 6 experiments—empirical and observational. design principles: Empirical experiments always involve two ♦♦ Replication—reduces error. or more treatments and have as their goal ♦♦ Randomization—prevents bias. the making of one or more comparisons. Empirical experiments can produce ♦♦ Blocking—increases precision. ambiguous results unless they are properly ♦♦ Control—permits assessment of change. designed. (See Experimental Design). ♦♦ Design—builds in the analytical method. Observational experiments involve making measurements at one or more points in These principles will be explored in the space or time. Space or time is the only following sections. “experimental” variable or “treatment.” Observational experiments can produce Determining objectives uninterpretable results unless the sampling method, scheme and strategy are carefully Before you conduct a field experiment with designed. (See Sampling Design.) pesticides, answer the following questions: Each type of experiment is planned ♦♦ What are the objectives? (What do you differently because different analytical want to prove?) methods are used to interpret the data. ♦♦ Observational experiments try to establish What is the design? (How are the relationships by rejecting the hypothesis treatments, plots, replications and that no relationship exists. Empirical controls arranged?) experiments try to establish comparative ♦♦ What sources of variation are there differences between treatments by rejecting within either the experiment or the the hypothesis that no differences exist. plot area? (Are there soil or varietal differences?) ♦♦ How many replications are needed? ♦♦ What is the sampling procedure? (Number of weeds per square foot, yield, etc.) ♦♦ When and how will the data be taken? ♦♦ How will the data be analyzed? ♦♦ How will the results of the experiment be used? (Publication, sales or demonstration?) The last question is very important because the way you intend to use the data from the experiment greatly affects the answers to the other questions. For example, an experiment designed to show that a pesticide increases yield will involve a different experimental plot design (a greater number of replications) and a different sampling procedure than an experiment designed to Figure 5. Example results from an empirical experiment to determine the illustrate the possible use of a pesticide in differences in delayed corn emergence produced by different corn herbicides. combination with a new tillage practice. 34 | Demonstration and Research Pest Control

Determining the analytical Steps of Experimentation method 1. Define the problem and objectives Observational experiments usually look for (hypothesis) predictable levels of an effect at differing levels of a single variable (time, space, etc.). 2. Plan experiments to test the hypothesis The design of observational experiments relies almost exclusively on “sampling • Select treatments design.” • Select experimental units Empirical experiments usually look for differences between two or more treatments. • Account for variability/experimental error The design of empirical experiments relies almost exclusively on “experimental design.” • Select experimental design Good experimental technique goes a long 3. Consider collection of data way toward minimizing error and bias. Every effort should be made to eliminate 4. Analyze data and interpret results these problems through appropriate experimental designs. The characteristics of good experiments are: the range of validity of an experiment. In a factorial experiment, the effects of one ♦♦ Simplicity. factor are evaluated under varying levels of a ♦♦ Degree of precision. second factor. ♦ ♦ Absence of systematic error. Calculation of degree of uncertainty ♦♦ Range of validity of conclusions. In any experiment there is always some ♦♦ Calculation of degree of uncertainty. degree of uncertainty about the validity of the outcome or conclusions. The Simplicity experiment should be designed so that it is possible to calculate the probability that The selection of treatments and the the results could have occurred by chance experimental arrangement should be as alone. It is important to realize, however, simple as possible, consistent with the that no answer is absolute and that all objectives of the experiment. generalizations drawn from an experiment Degree of precision should be made with care. The probability should be high that the experiment will measure treatment differences with the degree of precision the researcher desires. This requires an appropriate design and sufficient replication. Absence of systematic error The experiment must be designed to ensure that experimental units receiving one treatment differ in no systematic way from those receiving another treatment so that an unbiased estimate of the effect of each treatment can be made. Range of validity of conclusions

Conclusions should have as wide a range of Figure 6. Example results from an observational experiment to determine the validity as possible. Replications increase the relationship of delayed corn emergence to the time after application of a corn range of validity of conclusions. A factorial herbicide to the soil. set of treatments is another way to increase 7 experimental design | 35

Experimental design

Experimental design is a planned it is the experimental unit that gets the 7 interference in the natural order of events treatment. by the researcher. Much of our scientific The experimental units or plots in which knowledge has come from carefully the treatment is not made are called the observing and measuring what happens controls or checks. Control plots should when a selected condition or a change be included in all experimental field (treatment) is introduced. work. Failure to include control plots Designing an experiment is an extremely or not including enough control plots important process because errors made in yields questionable results that are usually the design can invalidate the results of the unacceptable for publication and sales entire experiment. Experimental design promotion. Check plots should be selected also is important because the researcher with the same objectivity as other plots. The wants to do more than simply describe the same variables that may affect treatment outcome. He or she also wants to make plots also may affect control plots. Thus, inferences about what factor contributed the location of control plots within a field to or caused events, and to do so without should not be selected arbitrarily. Likewise, ambiguity. Thus, proper experimental design control animals should not be selected is critical for ruling out alternatives and arbitrarily but should represent a random producing clear results. sample of the test population. The most able statistician cannot validate conclusions from an improperly Selecting treatments designed experiment. Experiments and demonstrations that have simple designs The objective, or purpose, of the study will are generally more successful than those determine the treatments included in an with elaborate designs. Knowing the limits experiment. Write down the test objectives of your resources—financial, land area so you can precisely define what it is you and other resources—can help you plan a want to find out. A test may have more than successful experiment. If you have questions one objective, although multiple objectives about your experimental design or method, should be closely related and clearly defined get help from a qualified statistician before to distinguish one from another. starting your research. The selection of treatments is usually Replication and randomization are basic logical if you can define the purpose of the components of valid research experiments. study. You should include ALL treatments Replication decreases experimental error, necessary to address the experiment’s while randomization prevents bias and objective. For example, if the purpose of increases the validity of data collected. an experiment is to determine which of five insecticides is most effective, then the treatments will include all five of those Design Considerations insecticides and an untreated control. If the purpose is to determine if any of the five Experimental units and control plots insecticides works better than your current choice, then the treatments will include An experimental unit is the smallest unit the five insecticides plus the insecticide to which a treatment can be applied at you presently use and an untreated control. random. Every treatment should have Accurately stating the purpose of the test an equal chance of being assigned to any before the treatments are applied in the field experimental unit. This ensures a valid is critical. After the treatments have begun, and unbiased estimate of the experimental it will be too late to add other treatments error and treatment differences. Remember, to answer the question you really wanted to address. 36 | Demonstration and Research Pest Control

The selection of treatments and the experimental design get more complicated as the question you are trying to answer gets more complex. It is common to want to test in the same experiment two (or more) things that influence crop production. For example, you may want to test how a particular post-emergence herbicide influences the yield on five different wheat varieties. The specific questions addressed in this case are: 1. What effect does the herbicide have on wheat yield? 2. What effect do the varieties have on wheat yield? Researcher prepares experimental treatments. Treatment selection depends on the 3. objectives of the trial and should test the hypothesis objectively. Photo Courtesy Does the herbicide have the same of Laurie Gordon, ODA. effect on each variety; i.e., are there any interactions? It is often desirable to have both a positive The third question may not be as obvious and a negative control in an experiment. The as the first two, but it will always be asked negative control helps you determine if the or implied if you are testing two or more treatments being tested work better than factors in the same experiment. In this some minimal treatment (or no treatment). example, you have to determine the effect The standard treatment helps you determine of the herbicide on each wheat variety and if the treatments being tested work better then compare those effects to each other. than the current standard practice. You To do this, the treatment list must include may have several control treatments in an each variety without the herbicide and experiment if you currently have several each variety with the herbicide treatment viable options from which to choose. For (a total of 10 treatments). With this list of example, if you currently can choose either treatments, you can make the comparisons of two fungicides to control leafspot, you necessary to answer the three questions. This may wish to include them both as controls example employs a “factorial arrangement in your experiment when you test new of treatments” that will be discussed in more products. You do not have to include all detail in a later section. currently available options as controls for Treatment selection also includes additional the experiment, but you can. treatments needed to provide a relative measure of effect. Comparing the yield of Plot size five new wheat varieties does little good if you cannot tell how those yields compare A plot is the area to which an individual with the variety you already grow. You treatment is applied. It can be any size, should include at least one variety with including a single plant growing in a pot or which you are already familiar (often several acres of a field. However, a plot must called a “standard” treatment) to provide be large enough to be representative of a a relative measure of how well the new much larger area. Plots that are larger than varieties produce. If you wish to test a new necessary take up more space and require nematicide, you should include a treatment much more work. Plots that are too small with the currently used nematicide and a may make it impossible to accurately assess treatment with no nematicide as a basis the effects of treatments. for comparison. Without the proper When deciding how large your plots controls, you will not be able to say that should be, consider the equipment to be the new nematicide worked better than used in planting, harvesting and treatment; the currently used nematicide or even that the amount of space available for the the new nematicide worked better than experiment; the number of treatments; no nematicide! The questions you wish the and the biology of what you are studying. experiment to answer should indicate what Accommodating equipment and space treatments should be included as controls. concerns makes it easier to conduct the test. 7 experimental design | 37

Accommodating biological concerns reduces This information will help you determine the chance of overlooking differences the minimum plot size necessary to get among treatments. Equipment and space useful data from the experiment. considerations are usually easy to identify, To get an accurate measurement of the but biological considerations are not always effect of pest management treatments, the obvious. plot must be large enough to account for If your equipment can plant, harvest and the uneven initial distribution of the pest treat four rows at a time, then the logical (pathogen, insect, weed, etc.). In some areas plot width would be some multiple of four the pest may be present to begin with, while rows (4, 8, 12 rows, etc.). Using any other in others the pest may appear only after it width (such as six rows) would make it more has spread from its initial location. This is difficult to conduct the experiment. The plot very important for pests that spread very length is generally more flexible than plot slowly (such as most soil-borne organisms). width. If you plan to weigh the harvest from Some diseases and pests are highly mobile each plot, the scales you have may influence and spread very rapidly (such as many the length plots should be. If your scales insects). In an insect management trial, are designed to weigh hundreds of pounds, measuring the effect of a treatment can your plots should be large enough to provide be very difficult if your plots are too small a harvest weight that can be weighed because the insects that you see in the plot accurately on those scales, and increasing may have simply spread from the plot next the length of plots is an easy way to do that. to it. To minimize this problem, you can The length of your plots can be adjusted so increase plot size and then collect data that all of your plots (all replications of all from the middle section of the plot. For treatments) will fit into the area available example, you might have an eight-row plot for your test. If you have a large area for but collect data only from the middle four your test, space may not be an important rows. The rows from which you do not consideration. collect data are often referred to as “buffer To accommodate biological considerations, rows” because they buffer the effect of the you should answer two questions: neighboring plots. If you do not use buffer rows when they are needed, you may fail to 1. How large a plot is necessary to detect differences among treatments and observe the biological effect (disease incorrectly conclude that treatments were severity, insect damage, weed frequency, ineffective. Buffer rows are often used when nematode population levels, etc.) that it is uncertain whether or not treatments can you are studying? influence nearby rows. 2. How large a plot is needed to minimize A similar concept involves the use of border the influence of a treatment (chemical rows along the edges of your test area. application) on the plots next to it? There is often a significant “border effect” at the edge of a field, where plants may grow differently than plants not at the edge. Although you may be able to minimize this problem with blocking, it is often better to eliminate the problem by not using the rows at the edge of a field in your experiment. Once the plots are large enough to be representative of a much larger area, further increasing plot size will not significantly improve the accuracy of the results. For example, in an experiment testing fungicides for control of white mold in bush beans, a plot four rows wide by 100 feet long should be just as good as a plot eight rows wide by 400 feet long. Plots that are larger than necessary may increase the amount of work Experimental canola plots are delineated by flags marking treatments. Photo Courtesy of Cory Cooley, ODA. required for an experiment, but usually will not adversely affect the test results unless they are so large that the plots within a 38 | Demonstration and Research Pest Control

block are no longer uniform. Plots that number of infected sites. You would not are too small may prevent the accurate have been able to determine that if you had assessment of treatment effects. If you have used only one treated and one untreated limited space for an experiment, use more plant unless the number of spots on the replications to ensure accurate results. treated plant was much larger than the number on the untreated plant. But what Controlling Experimental Error if the untreated plant had 19 spots and the treated plant had 21 spots? That might have led you to conclude that the fungicide Replication did not work, or even that it increased susceptibility to the disease. Adequate Experimental error can result from the replication can minimize this problem. inherent variability among individual experimental units. All experiments produce With several replications of each treatment variable results. The variability within a it is common to have data like that in the field and from one field or farm to another rose example where the treatment means can significantly affect the outcome of are different but individual measurements experiments and demonstrations. The may overlap. In this example, the lowest inherent differences between individual measurement from the untreated plants animals, herds and flocks can do the same. was 19, and the highest measurement from You will not get the same results from two the fungicide-treated plants was 21, but the plots or two sets of animals that receive treatment means were 22.8 for the untreated identical treatments. In statistical terms, plants and 18.8 for the treated plants. this natural, uncontrolled variability among Replication of treatments increases your experimental units is called “experimental ability to detect differences in treatment error.” means. Having more replications allows you to identify (statistically) smaller differences To ensure that the differences among in treatment means than you could identify experimental units do not unduly influence with fewer replications. an experiment, the experiment is repeated, or replicated, several times in different The number of replications you need is locations. By replicating an experiment, influenced by the biology of what you are researchers are able to estimate the amount testing, how close together the treatment of experimental error in the study and means are, and how much variation there is make the results more precise. The number within a treatment. For field tests in plant of replications required in a particular pathology, nematology, weed science, soil experiment depends on the magnitude fertility and entomology, a minimum of of the differences the researcher wishes four replications is suggested; five or six to detect and the variability of the data. replications are much better. If treatment Carefully considering the number of means are close together or variation is replications needed at the beginning of the relatively large among the plots that received experiment can save much frustration later. the same treatment, then you may need more replications to detect differences Example among treatments. Just as the data may vary within a replicated You have ten rose bushes and you want to treatment, the results may vary among determine whether a new fungicide will experiments if the whole experiment is protect the bushes from black spot, a fungal repeated. This can happen because of leaf disease. You pick five plants to leave different weather conditions, the presence untreated as a control and spray the other of different diseases or insects, or many five with the fungicide. Later, when black other factors beyond your control. This spot is evident on the leaves, you count the does not mean that the results of a single number of diseased spots on each plant experiment are not valid, but it does make and compare the two treatments. The five it dangerous to draw conclusions from a untreated plants have 26, 21, 19, 25 and 23 single experiment. The one set of results you infected spots, giving an average (or mean) have might not appear again if you repeated of 22.8 spots per plant. Treated plants have the test several times. If the test were 20, 15, 18, 21 and 20 spots, giving a mean repeated with identical treatments and you of 18.8 spots per plant. Statistical analysis got similar results, then you could be much shows that the fungicide did reduce the 7 experimental design | 39

more confident that your conclusions were Unrandomized Example 1 correct. An experimenter wants to test five corn You can minimize the effect of variation hybrids (labeled 1 through 5) and plants the if it has an identifiable cause, but there hybrids in the same order in each block: 1, 2, will always be some variation among 3, 4, 5 (Fig. 8). If hybrid 2 is naturally much experimental units that you can not taller than the others, it can slightly shade control. The purpose of replication is to the hybrids planted next to it (varieties 1 allow you to more accurately estimate and 3). This unfairly makes them perform a how each treatment performed in spite of bit worse than they would if they were not uncontrolled variation in the experiment. planted in the shade of hybrid 2. This unfair advantage given to one experimental unit Randomization over another is called “bias.” In this case, the experimental design is said to be “biased” Another component of good experimental against hybrids 1 and 3. design is randomization (Fig. 7). Treatments are randomized to ensure that the variability Unrandomized within an experiment occurs purely by chance. Every treatment should have an 1 2 3 4 5 equal chance of being assigned to any experimental unit. This ensures a valid and 1 2 3 4 5 unbiased estimate of the differences between treatments. The assignment of treatments to 1 2 3 4 5 plots in a purely objective (random) way is called randomization. 1 2 3 4 5 Randomization in an experiment means 1 2 3 4 5 that in both the allocation of experimental units and the order in which treatments are Figure 8. Unrandomized corn hybrid trial where variety applied, any experimental unit has an equal number 2 is taller than the other varieties and shades chance of being selected for any particular adjacent rows. The effects of shading decrease with treatment. That is, treatments are assigned distance on either side of the row containing variety number 2. to plots with no discernible pattern to the assignments. Or, if a pattern is discernible, it is the result of pure chance. Proper randomization averages out the effects Unrandomized Example 2 of extraneous factors that may produce An experimenter plants four alfalfa varieties differences between experimental units. (labeled A through D) in the same order Randomization also removes researcher across each plot (Fig. 9). In the field bias. For statistical analysis to be valid, where they are planted, soil fertility and evaluation methods require observations productivity get progressively lower as you that are independently distributed random go from one side of the field to the other. variables. The reason randomization is Within a block, variety A is always planted so important is that the order in which in more fertile soil than any of the other treatments are selected for an area or a variety, while variety D always grows in the treated population may affect the outcome. least fertile soil. The lack of randomization Unrandomized Randomized always gives variety A an unfair advantage in the trial. A A A A A B D D Unrandomized Example 3 B B B B D C C A Researchers want to test four different feed mixes on feeder calves. They have 16 calves C C C C C D B C available for the experiment, so each feed will be given to four calves. The 16 calves D D D D A A B B arrive and are placed in a large holding compound until they are ready for the Figure 7. Test plots on the left are not randomized. Plots on experiment. The researchers assume that the right are randomized. The letters (A-D) represent the four calves will be caught “at random.” They enter treatments in this test. the compound, catch four calves, and assign 40 | Demonstration and Research Pest Control

Unrandomized Lottery method—pulling numbers out of a hat A B C D The simplest way is literally to pull the A B C D numbers out of a hat. Assign each treatment a number, write the numbers on individual A B C D pieces of paper, mix the slips of paper up, and then select the slips one at a time A B C D without looking at them first. The order in which the numbers are drawn is the order High Low in which they will be arranged in a block. fertility fertility Repeat these steps for each block in the experiment. Figure 9. Unrandomized alfalfa variety trial in field with progressively lower fertility west to east. The unrandomized design favors alfalfa variety A over all Random number tables the other varieties, while variety D gets the worst of it in every replication. Another way to select experimental units or assign treatments is to use a table of random numbers. A random number table them to treatment A. They catch the next is a table of digits. The digit in each position four calves and assign them to treatment B. in the table was originally chosen randomly They follow this process through treatments from the digits 1, 2, 3, 4, 5, 6, 7, 8, 9 and 0 C and D. by a random process in which each digit is Although the experiment appears to be equally likely to be chosen. A number itself randomized, it isn’t. The first calves caught cannot be random except in the sense of could be the slowest, weakest, most unthrifty how it was generated. Generating a random and least likely to respond favorably to any number means that all elements of the set of the treatments. The last ones caught could were equally probable as outcomes. This be the quickest, healthiest, thriftiest and is equivalent to statistical independence. most likely to respond favorably to any kind Tables of random numbers are available in of feed. This would bias the results in favor most statistics text books. For reference, of the last calves caught. If the experimental a random number table is provided in results came out to the disadvantage of Appendix C. treatment D, there would be no way to The first step in using a table of random determine if the results were a consequence numbers is figuring out how big a number is of treatment D or the fact that the strongest, required. This will depend upon the number thriftiest calves were placed on that of experimental units and the number of treatment by the selection process. replications. Do not keep using the same In all three of the examples, randomization part of a random number table over and could have prevented the unintentional over again. Change starting points and bias because the arrangement of the directions. Pick a starting point at random, treatments would have been different either by using the “look away and stick a among the experimental units. Because pin in it” method or by drawing starting row you cannot anticipate all the influences and column numbers from a hat. that may introduce bias into a test, ALL After picking the starting point, sets of experiments should be randomized. numbers are gathered from the table either by reading left, right up or down from the Randomization methods starting number. For each value desired, pick the number of digits to match the highest There are many ways to randomize samples, number and ignore any values picked that treatments and experimental units. In are higher than that value. For example, to selecting numbers at random, it is not get 10 random numbers between 1 and 60: so much the method of producing the 1) randomly select a start point in the table, numbers that matters but the properties of 2) select this digit plus the one next to it, 3) the numbers produced. They should have move up, down, left or right, 4) choose the the properties we would expect “random” next two digits. Repeat this process enough numbers to have. times to get ten two-digit numbers. If you 7 experimental design | 41

encounter a value greater than 60, simply The advantages of the completely ignore it and select another. randomized design are the ease of set up and ease of analysis. Disadvantages include Computer-generated random numbers a loss in precision in determining differences among the whole-plot treatments and A third way to randomize experimental difficulty in making applications to units and treatments is to use pseudo- experiments with large numbers of random numbers. treatments. Pseudo-random numbers. A pseudo- A particularly advantageous feature of random number is a number belonging to the CRD is that it permits the analysis of a long sequence generated by an computer unbalanced data. This means that unequal that appears to be random but eventually sample sizes may be analyzed. If a plot repeats itself exactly. That is why numbers is lost to hail, fire, flood or other natural in the sequence are called “pseudo- disaster, the data can still be analyzed. If random” numbers. In the short run, these a cooperator inadvertently sprays one of sequences of pseudo-random numbers have the check plots or fails to treat one of the an apparent randomness. These pseudo- treatment plots, these may be discarded and random numbers are sometimes random the data from the remaining plots will still enough if you don’t have millions of units to be usable. This is not possible with blocked randomize. experiments, which require balanced data True random numbers. Modern computers for valid analysis. do a better job of approaching true Although completely randomized designs randomness than early computers. More are flexible and simple, estimating sophisticated programs are available to experimental error with this design may generate better (less predictable) pseudo- be less precise than with other designs. random numbers. Some web sites actually The completely randomized design is claim to produce batches of true random not usually the most efficient design for numbers from a hardware-based random research in field crops and may be more number generator. An example of such a appropriate for trials with livestock. The web site is http://www.random.org. This completely randomized design is also used web site also includes a discussion on frequently in greenhouse tests, though pseudo vs true random numbers. blocking is often useful even in the more Types of experimental design controlled environment of a greenhouse. An example of completely randomized design is shown in Figure 10. Here, three 1. Completely randomized design (CRD) herbicides + one control = four treatments x three replications = twelve plots. The The completely randomized design is treatments are assigned to the plots at the simplest experimental design. In this random. design, treatments are replicated but not blocked, which means that the treatments are assigned to plots in a completely random manner. This design is appropriate if the A A C C entire test area is homogeneous (uniform in every way that can influence the results). A B C B A CRD is set up by assigning treatments and controls at random to a previously B D B C determined set of experimental units. Any number of treatments may be tested in D D D A this design. It is desirable to assign the same number of experimental units to each treatment and control, but it is not essential. Figure 10. Completely randomized design with three When plots are laid out within a field, replications, three herbicide treatments (A, B, and C), the number of plots is determined by and one control or “check” (D). multiplying the combined number of treatments and controls by the number of replications desired. 42 | Demonstration and Research Pest Control 2. Randomized complete block (RCB) A B C A A B A C The randomized complete block design (Fig. 11) is used to account for natural variability that would otherwise obscure treatment differences. In this design, the A B C B B C C A treatments are assigned at random to a group of plots called a block; a block is a grouping of single occurrences of each treatment. Because adjacent plots are more C A B C C A B B likely to produce similar yields or have similar pest infestations or similar fertility Block 1 Block 2 Block 3 Block 4 than those separated by some distance, the block is kept as compact as possible. This is Figure 12. The shaded area represents an area of the field that is different from accomplished by placing the plots, usually the unshaded area. Treatments A,B and C are replicated but not blocked in the field on the left (CRD). On the right, treatments are replicated and blocked; long and narrow, close together. The number each block contains one plot of each treatment (RCB). of treatments also should be as small as possible. error). This allows the statistical analysis The randomized complete block design to identify treatment differences that is the most commonly used design in would otherwise be obscured by too much agricultural field research. In this design, unexplained variation in the experiment. treatments are both replicated and blocked, Variation in an experiment is of two which means that plots are arranged into types—variation for which you can account blocks and then treatments are assigned to in the statistical analysis and variation that plots within a block in a random manner is unexplained. The goal in blocking is to (as in the right side of Fig. 12). This design allow you to measure the variation among is most effective if you can identify the blocks and then remove that variation from patterns of non-uniformity in a field, such the statistical comparison of treatment as changing soil types, drainage patterns, means. If you can anticipate causes of fertility gradients, direction of insect variation, you can block the treatments to migration into a field, etc. If you cannot minimize variation within each block and identify the potential sources of variation, remove some variation from the statistical you should still use this design for field analysis. The mathematics of how blocking research but make your blocks as square as allows you to reduce unexplained variation is possible. This usually will keep plots within beyond the scope of this manual. a block as uniform as possible even if you cannot predict the variation among plots. In the most common experimental designs, a block will contain one plot of Blocking refers to physically grouping each treatment in the experiment. If an treatments together in an experiment to experiment has five treatments, then each minimize unexplained variation in the data block will contain five plots, with each plot you collect (referred to as experimental receiving a different treatment. When a block contains one plot of each treatment, Blocks then each block represents one replication 1 2 3 4 of each treatment. For this reason, blocks are frequently referred to as “replications” A B C C or “reps,” but the concept of blocking should not be confused with the concept of B D D A replication; replication and blocking serve different purposes. In agricultural research, C C B D Low fertiity Low High Fertility High D A A B A B C B C A A C B

Low High Block 1 Block 2 Block 3 Figure 11. Randomized complete block design replicated four times with three herbicide treatments Figure 13. An easy way to arrange blocks is to put them side by side across the (A, B and C) and one untreated control (D). field. Letters represent different treatments. 7 experimental design | 43

field plots are almost always blocked even is what makes blocking effective. Your goal when no obvious differences are present in is to maximize the differences among blocks the field. It is much better to block when while minimizing the differences within a you did not really need to than not to block block. when you should have blocked. The shape of blocks is not important as long Blocking is a very powerful tool that is most as the plots within a block are as uniform effective if you can anticipate sources of as possible. Ideally, the only differences variation before you begin an experiment. among plots within a block should be the For example, in a herbicide trial, one side treatments. Blocks in field experiments are of a field may have a history of more usually square or rectangular, but they may severe weed problems than another. If you be any shape. Blocks in the same experiment just scattered your treatments randomly do not have to be the same shape; the shape through the field, a lot of the variation in of individual blocks will be determined by the data you collected could be due to the variations in the field that you are trying to increased weed pressure on one side of the minimize. If you are not sure what shape field. Such variation would make it difficult your blocks should be, square or nearly to determine how well each treatment square blocks are usually a safe choice. worked. Because you know one side of the Blocks may be arranged through the field in field will have more weeds, you can remove many ways. If the field is wide enough, an that source of variation from the statistical easy way to arrange blocks is to place them analysis by blocking and improve your side-by-side all the way down the field. But chances of identifying differences among blocks do not have to be contiguous and treatments. may be scattered through the field in any The process of blocking follows a logical way that is convenient for you. sequence. First, you determine that there is Note that each treatment occurs only once something (weeds, drainage, sun/shadow, in each of the four blocks. Treatments are water, soil type, etc.) that is not uniform assigned at random to plots within each throughout the experimental area (field, block, with a separate randomization made greenhouse, etc.) that may influence for each block. Crop rows should run whatever you are measuring (yield, plant perpendicular to the fertility gradient to height, etc.). Then you arrange your minimize experimental error. treatments into blocks so that the area within each block is as uniform as possible (see Fig.14). Though the area within a block 3. Split plot should be relatively uniform, there may be large differences among the blocks; but that A split-plot experimental design is a special design that is sometimes used CR Design RCB Design with factorial arrangements of treatments. 1 2 3 4 An example would be an experiment to evaluate both pesticide performance and A A C C A B C C crop management practices (e.g., tillage, row spacing, crop variety), such as the A B C B B D D A effectiveness of three herbicide treatments in no-till and conventional tillage (Fig. 15). To B D B C C C B D simplify the experiment, tillage treatments are established as whole plots. Each whole D D D A D A A B plot is divided into four subplots and the herbicide treatments (three herbicide Low High Low High treatments plus a control) are randomized fertility fertility fertility fertility within each whole plot. The split plot design also can be used Figure 14. The elimination of bias through blocking: an herbicide trial with three treatments (A, B, and C) and an when some constraint prevents you untreated control (D) replicated four times in a field with from randomizing the treatments into a a gradient of increasing fertility from west to east to be randomized complete block design. Such a evaluated by its effect on both target weeds and on crop constraint might be equipment limitations growth response. Note how the inherent bias favoring increased growth response in treatments A and D in the or biological considerations. completely randomized design (CRD) is eliminated by blocking in the randomized complete block design (RCBD). For example, the equipment you have may make it difficult to put out a soil fumigant 44 | Demonstration and Research Pest Control

in randomized complete blocks, but you nt nt ct ct ct ct may be able to put out the fumigant if H1 H2 C H1 H1 H3 all treatments within a block that get the fumigant are clustered together rather nt nt ct ct ct ct than scattered throughout the block. You H3 C H2 H3 C H2 can use a split-plot experimental design nt nt nt nt ct ct to work around this limitation as long H1 C H3 C C H1 as you are able to randomize the other nt nt nt nt ct ct factors. There are other situations when this H2 H3 H1 H2 H2 H3 design is appropriate, but a constraint on randomization is the most likely to occur. Figure 15. Randomized split plot design. No-till (nt) and conventional tillage (ct) are the whole-plot treatments and herbicide treatments (H1, H2 and H3) Example are subplot treatments. C is a control treatment in the subplot treatments. Suppose you want to test the effect of five fungicides to control stem rust on block design. The subplot treatments can two varieties of perennial ryegrass. In then be randomized within each whole plot this test, you would have a 2 x 5 factorial treatment (see Fig. 16). arrangement of treatments: The two factors The advantage of the split-plot design is would be varieties (two levels of this that it simplifies experiments where large factor) and fungicides (five levels of this equipment is used. factor). Because a factorial arrangement of treatments is not an experimental design, 4. Split block you still have to select an experimental design that best meets your needs. The split-block design is a variation of the If you are able to randomize varieties and split-plot design. Subunit treatments are fungicides within a block, then you should applied in strips across an entire replication pick a randomized complete block design. of main plot treatments (Fig. 17). This If there is some reason why you cannot arrangement often facilitates physical completely randomize the treatments operations in the subunits but sacrifices within each block, then you may be able to precision in comparing the effects of the use a split-plot design to work around that subunit treatments. limitation (as in the lower part of Figure 16). In a split block design, two sets of For example, you may have a six-row planter treatments are randomized across each but only enough space in the field to put other in strips in an otherwise RCB out four-row plots. To resolve this dilemma, design. It is used where logistics make it you could plant all of the plots that have necessary to run treatments completely the same peanut variety together within a across each block. The number of blocks block and then randomize the five fungicide treatments within each peanut variety. 3 5 1 4 2 1 4 3 2 5 Block 1 In split-plot designs, the terms “whole A B B B A A A B B A plots” and “sub-plots” refer to the plots 2 5 4 2 4 3 1 1 3 5 Block 2 RCBD into which the factors are randomized. A B B B A A A B B A As the names imply, whole plots are 1 3 4 5 3 4 2 2 1 5 Block 3 subdivided into subplots. In Figure 16, a A B B B A A A B B A whole plot would be the areas designated 5 2 1 4 3 1 3 5 4 2 Plot 1 with A or B, and the subplots the areas A A A A A B B B B B designated 1, 2, 3, 4 or 5. In this example, 5 3 1 2 4 4 3 2 1 5 Plot 2 Split A and B could represent two varieties (two B B B B B A A A A A plot levels of one factor) and the numbers could 4 3 5 1 2 2 1 3 5 4 represent different fungicides (five levels of A A A A A B B B B B Plot 3 a second factor). Each whole plot serves as a block for the subplot treatments. Figure 16. A 2x5 factorial arrangement of treatments in a randomized complete To assign treatments in a split-plot design, block design (above) and in a split-plot design (below). A and B represent two levels of one factor, and the numbers (1-5) represent five levels of a second start by identifying where each block factor. The combinations (e.g., 4A, 5B, etc.) denote individual treatment will be. The whole plot treatments will combinations. Either experimental design could be used, but the randomized be the treatments that you are unable to complete block design is preferred unless the split-plot design is required by randomize into a randomized complete some limitation on randomization. 7 experimental design | 45

Block 1 Block 2 Block 3 A factorial arrangement of treatments is A1 C2 C3 Bc B1 C3 Bc A2 Cc A2 B1 C3 not an experimental design, though you B1 B2 A3 Cc A1 B3 Cc B2 Ac B2 A1 B3 will often hear it referred to as a factorial design or a factorial experiment. A factorial C1 A2 B3 Ac C1 A3 Ac C2 Bc C2 C1 A3 arrangement of treatments means that the experiment is testing two or more factors Figure 17. Split-block design with three replications, three main treatments (A, B, and C) of corn planting dates, three subplots (1, 2, and 3) of fertilizer at the same time, and that the experiment treatments, and a control treatment (c). includes all combinations of all factors. The term “factor” is used to describe a group of is the number of replications. This design treatments that have something in common. is useful in orchards and vineyards where Fungicides, sources of nitrogen, or corn pesticide applications are made with air blast hybrids could be considered factors in an sprayers. However, it can be used anywhere experiment. Factors may be defined broadly that treatments have to run completely or narrowly in different experiments. All across each block. herbicides may be grouped as a factor in one experiment, but preplant and postplant herbicides may be treated as separate factors 5. Latin square in another experiment. A single-factor experiment tests one factor at a time; a two- The Latin square design groups treatments factor experiment tests two factors at once. in two different ways—by columns and rows (Fig.18). Every treatment occurs Most simple on-farm experiments are once in each block (row) and once in each single-factor experiments (in a Completely column. Variability across the experimental Randomized or Randomized Complete area is measured and removed in two Block design) and compare things such directions. With the Latin square design, as crop varieties or herbicides. But it is the number of treatments must equal sometimes useful to test two or more the number of replications. With a large factors at once. For example, a two-factor number of treatments this design becomes experiment would allow you to compare the cumbersome. Usually, this design is used yields of five corn hybrids at three planting in small experiments where there are four dates. This accomplishes three things at to eight treatments. Consult the appendix once: of this manual for more examples of Latin 1. It allows you to compare the corn square plot design. hybrids with each other. 2. It allows you to evaluate the effect of A B C D E planting date. 3. It allows you to determine whether B D E A C varying the planting date changes the C A D E B relative performance of the hybrids (e.g., one hybrid may perform well only if D E B C A planted early).

E C A B D The first two could be done in separate single-factor experiments, but the third can be achieved only by having both factors in a single experiment. This becomes Figure 18. Latin square design with five replications of four herbicide treatments (A, B, C, D) and a control especially important if one factor can have (E). a significant influence on the effect of the other factor. For example, you might test soybean varieties as one factor and Factorial arrangement nematicides as another factor. If a few varieties have good nematode resistance A factorial arrangement is useful when but others do not, they may appear equally investigating the effects of each of a number good when effective nematicides are used of variables, or factors, on some response. but show significant differences when There are advantages to be gained by nematicides are not used. In cases like this, combining the study of several factors in the the effect of one factor (variety) is strongly same experiment. influenced by the other factor (nematicide). 46 | Demonstration and Research Pest Control

When one factor influences the effect or a randomized complete block design. of the other factor, there is said to be a The top half of Figure 16 shows a factorial significant interaction between the two arrangement of treatments in a randomized factors. It can be very important to know Rep. 1 if there is an interaction between factors. a b a b a b If there is an interaction, you can make 0 0 0 1 0 2

predictions or recommendations based on a1b0 a1b1 a1b2 the results of single-factor experiments a b a b a b ONLY when all other factors are at the 2 0 2 1 2 2 same levels as in the experiment. If you Rep. 2

change some factor not included in the a0b1 a1b1 a2b2 experiment, the results from your single- a b a b a b factor experiment may no longer be valid. 0 0 1 2 0 2 a b a b a b With a factorial arrangement of 2 1 2 0 1 0 treatments, all values (or levels) of each Rep. 3 factor must be paired with all levels of the a b a b a b other factors. If you have two nematicides 1 0 2 1 1 2 a b a b a b and five soybean varieties, then your 1 1 2 2 2 0

treatment list must include each variety a0b0 a0b1 a0b2 with each nematicide for a total of ten treatments. This would be referred to as a Figure 19. Factorial design with three replications and “2x5 factorial” to denote how many factors two herbicide treatments (a and b) each applied at two were present in the experiment and how rates (1 and 2) with controls (0). The untreated controls are considered treatments, so this is a 3x3 factorial (9 total many levels of each factor were used. The treatments) with 3 replications for a total of 27 plots. number of treatments increases quickly when you add more levels for a factor (if you used three nematicides instead of two, complete block design. you would have 15 treatments instead of ten). So choose your levels carefully or the Factorial experiments are highly efficient experiment can get too large to manage. because every observation supplies information about all the factors included A factorial arrangement of treatments can be in the experiment. Also, these types of a very powerful tool, but because the number experiments investigate the relationships of treatments can get very large it is best between the effects of different factors. The used when you believe that the factors may treatments consist of all combinations that influence each other and have a significant can be formed from the different factors. interaction. If there is no suspicion that Each treatment combination is randomized the factors may influence each other, it is within each replication. Such a design is often easier and more thorough to test the useful when studying the effects of several factors in separate experiments. A factorial rates of several pesticides applied in the arrangement of treatments can be used with a same experiment (Fig. 19). completely randomized experimental design 7 experimental design | 47 Summarizing experimental design Experimental design checklist 3 Determine the objective of the The following checklist can be used in test. designing an experiment. These items may 3 be addressed in any order. Select treatments that address the objective. Consider Good experimental technique goes a long including positive and negative way toward minimizing error and bias. controls. Every effort should be made to eliminate 3 these problems through appropriate Determine what data should be experimental designs. To help eliminate collected, and when it should experimental error and bias: be collected, to address the ♦ objective. ♦ Apply all treatments uniformly. 3 ♦ Select the number of ♦ Measure all treatment effects in an replications to use. Consider unbiased way. four replications a minimum. ♦ 3 ♦ Prevent gross errors. Determine how big individual ♦♦ Control external influences so that all plots will be. 3 treatments are affected equally. Select an experimental design. 3 Properly designing and implementing a field Determine how blocks should trial may seem complex the first time, but be arranged in the field. it is really a logical process that should not 3 be intimidating. You may need help the first Randomize treatments within time you design a trial to ensure that you are blocks. not overlooking something important, but if you learn the principles involved in the process, you should quickly gain confidence in your ability to conduct experiments on your own. 48 | Demonstration and Research Pest Control 8 Data collection | 49

Data collection

You could collect an almost infinite amount biology of the organisms involved and 8of data in any experiment, but not all how your data addresses the test objective, of it will be useful. Proper planning will then you should be able to tell if you are ensure that you collect the right data to collecting enough data. address your test’s objective. The “right” You should take photographs of any data to collect usually can be determined differences among treatments that are easily by examining the stated purposes of the visible. To most farmers, a picture is more experiment. For example, if the objective convincing than a graph or data table. of a peanut leafspot fungicide trial is “to evaluate the ability of five fungicides to reduce leafspot incidence and Bias severity,” then collecting data on leafspot incidence and severity Data collected from a sample that is not and peanut yield should seem representative of the population will not obvious. Collecting data on rainfall accurately reflect one or more population and temperature, which strongly characteristics. These are called “biased influence leafspot on sugarbeets, samples.” A famous example of a biased may be worthwhile because it can sample occurred in the 1936 presidential help you explain your results. But election polls. One large poll (2,000,000 collecting data on the physical people) predicted that Landon would properties of the soil does not seem defeat Roosevelt. A smaller poll (300,000) to be related to the objective. It is predicted that Roosevelt would win. The useful to ask yourself, “How can this smaller poll, taken from U.S. census lists, data be used?” If you have trouble was correct because of its unbiased sample. Sampling for insects in an The larger poll had drawn its sample from experiment. Photo source: answering that question, then UC Davis. collecting the data may be a waste telephone directory lists of middle- and of time. It is much more common upper-income citizens, most of whom voted for people to collect too little data for Landon. Thus, the larger sample was than to collect too much data. Consult the biased toward more affluent voters because section on Sampling Design. those without telephones had no chance of being chosen. A biased sample is one in Deciding what data to collect is only part of which not all members of a population are the process. You also have to decide when equally likely to be chosen. Another kind to collect that data and whether you need to of bias is called “statistical noise.” It is the collect the same type of data on more than inherent variability from one experimental one occasion. For example, in a nematicide unit to another. Problems with “statistical trial, it is not sufficient to collect nematode noise” can be lessened by enlarging the population data at harvest; you must also sample. However, a biased sample will collect data at planting to ensure that the always produce biased data. plots started out equal. It is usually a good idea to collect nematode population data Collecting unbiased data in the middle of the season also because even with effective treatments nematode It is critical to collect unbiased data. The populations can sometimes increase to the only way to ensure this is to collect data level of the untreated control by the end of without knowing what the treatment was in the season. The biology of the organisms that plot. That would be difficult to do if the involved will determine when and how treatment were written on a stake in front of frequently data should be collected. each plot. Instead, use some type of code on So, how much data is enough? The answer is the plot stakes so that you have to decode that you want enough data to fully address the stake number to determine what the the test’s objective. If you understand the treatment was. You can make up any code 50 | Demonstration and Research Pest Control

animals, the yield of crops, the amount of damage, or anything else. Because all experimental units are different, what is being measured is called a “variable.” Value Each measurement recorded for a variable (e.g., the number, height, weight, yield, amount, etc.) is called a “value.” Population A population is a set of elements about Figure 20. Sampling example: Potted ornamental which a researcher wants to make inferences. plants are treated with seven fertilizer levels and two Elements may consist of people, plants, insecticide rates. The design includes14 replications for animals, objects, etc. Researchers draw a total of 196 pots. This example shows a non-random sample of 20 taken from the upper left corner of the conclusions about an entire population field. Effects due to influences in this region of the field from inferences made during observations (such as more light) could influence the measurements of some population characteristic. The using this sampling technique. size of a population is the number of observations possible in it. Sometimes you like as long as the person collecting the a population is too large or difficult to data cannot tell from the plot stake what the observe in its entirety, so a portion of the treatment was. For example, you can number the plots sequentially (1, 2, 3, etc.) and have a sheet of paper listing what treatment was applied to plot 1, plot 2, etc. When you collect the data, you write down your observation for plot 1 and later look at your list to see what treatment was in that plot. If you know what treatment was in a plot, or which plots were the untreated controls, your evaluations (disease severity ratings, insect damage ratings, etc.) may inadvertently be influenced. Your subconscious may slightly increase the ratings for untreated plots and decrease it for the plots with treatments that you think should work well. You will probably not even be aware that it is happening, but these subtle influences can change the data enough to affect your ultimate conclusions from the test. If you do not collect unbiased data, you cannot be certain that your conclusions are correct. Sampling Design To understand sampling, you must understand the following terms: variable, Figure 21. Types of population distributions: a) random distribution often value, population, sample, subsample and characteristic of populations in which there is little interaction between members, such as dandelions and other weeds; b) uniform distribution (sometimes bias. called regular) characteristic of populations in which there is strong territorial interaction or where human activity produces regular spacing, such as orchards, Variable vineyards and field or row crops; c) clumped distribution (sometimes called contagious) characteristic of populations in an environment with uneven distribution of resources (examples: aphids, mites diseases). Sampling designs Research measures some attribute of the take into account the possibility of non-homogenous environments and varying experimental unit, such as the size of cells, types of population distributions. the number of organisms, the weight of 8 Data collection | 51

demand that a sample be selected in such a way that it won’t present an incorrect or biased view of the population. If statistical inference is to be used, there must be a way of assigning known probabilities of selection to each sample. If the sample is selected in such a way that each member of the set has an equal probability of being selected, the sample is called a “random sample” (Fig 22). Subsamples Subsampling is the recording of more than one measurement on the same Figure 22. Random sample taken from potted experimental unit (Fig. 23). Subsampling is ornamental plants (see fig. 20) in which sampling often desirable in recording the effects of a units have been numbered from 1 to 196. Twenty samples have been selected by choosing sampling treatment in an experiment. For example, units at random, and each sampling unit has an equal when entomologists sample ten corn plants probability of being selected. per plot (where the plot contains more than ten corn plants) to estimate resistance to corn borers, each of these plants is a whole population is observed. This set of subsample. To tell the difference between observations is called a “sample.” a subsample and a replication, remember Population distributions have three general that an experimental unit is defined as the types of dispersal patterns: random, uniform smallest unit to which a treatment can be (sometimes called spaced) and clumped applied at random (meaning that each unit (sometimes called contagious) (Fig. 21). is chosen independently of any other unit). Random sampling designs allow researchers to select members of the population for Example 1 sampling with equal probability Assume you have 16 pots of greenhouse- grown corn to use in testing three Samples insecticides and an untreated control. Let’s assume you choose the four pots on A sample is a small part of a population the nearest table and sprayed them with intended to be representative of the whole. an insecticide. They would become one In research, a sample is a subset of an experimental unit, not four replications entire population or process, the elements of a treatment. To be replications, the of which are selected in an intentional and pots needed to be chosen at random predetermined way. Scientific standards from among the 16. Thus, there are four possible sampling units in this experiment. If the weight of the plant was the desired measurement, up to four subsamples could be taken for each experimental unit. Example 2

In an irrigation trial, if you use a water regime (like 75% ET) on one span under a center pivot, then all the plants under that span all the way around the circle belong to the same experimental unit. You can randomize a center pivot irrigation trial (where treatments vary by inches of Figure 23. Subsampling: 20 random samples with subsampling within each water applied) by randomly deciding which sample. The 15th sample (Sample 133) shows ten randomly selected subsamples treatment to put under which nozzles, but marked with Xs within it. In our example of potted ornamental plants (Figs 20 all plants under the same nozzles will always and 22), the subsample might be ten leaves selected at random from the plant. belong to the same experimental unit. Although the sample contains ten subsamples, the sample itself counts as only one sample. However, in a trial for something else, such as a herbicide, which uses pivot irrigation 52 | Demonstration and Research Pest Control

but where irrigation is not a formal factor, Simple random sample plots under the same nozzles are not necessarily in the same experimental unit. This is the most widely used method. Here, differences in irrigation will add to the A simple random sample uses various experimental error. methods of randomization to ensure that each element in an experimental unit has an Example 3 equal probability of being selected. Crop variety tests are often planted in strips, with one strip per variety. In this case, a strip is an experimental unit. All samples taken from a particular strip are subsamples. Taking many samples from the same experimental unit is not replication. Types of sampling Census A census counts or records values for the entire population. A census is possible only when the entire population and all individual elements within the population Figure 24. A simple random sample taken from a non- are accessible. homogenous field. Probabilistic sampling Stratified random sample If the purpose of your research is to draw conclusions or make predictions In this method, the population is first about a population as a whole (as most divided into two or more mutually exclusive research is), you must use a probabilistic groups based on some variables of interest sampling approach. The key to this type in the research. That is, the population is of sampling is random selection. But do organized into homogenous subsets before not confuse random selection with random sampling. Then random samples are drawn assignment. Both involve randomization, from each subset. This method ensures but their purposes are different. Random that elements from smaller strata (non- selection is how you draw the sample (see overlapping subpopulations) are included Randomization). Random assignment is in sufficient numbers to allow comparison. how you assign treatments to experimental Samples are drawn from each of the units for experimental or control purposes different strata. (see Experimental Design). Random sampling designs use various selection methods. They all provide an equal probability of selection for every member of the population, but they are used in different situations depending on whether the population is distributed in a random manner, a uniform manner (sometimes called spaced) or a clumped manner (sometimes called contagious). Sampling design Below are some common ways random sampling is designed. These principles can be applied to sampling entire experimental Figure 25. Stratified random sample with equal sample numbers taken from light (-) and dark (+) sections of units or to subsampling within an the field (sampling strata). experimental unit. 8 Data collection | 53

before the researcher examines the area. Tip: Don’t make extra work for yourself For example, there is no subjective decision when entering data. One experimental if you start at a random point in a field unit receives one value, although it may and sample every tenth potato plant for Colorado potato beetles. In essence, be the average of several subsamples. systematic sampling is random sampling Taking measurements on several elements with a system. (plants, animals, etc.) within the same Advantages of systematic sampling over experimental unit reduces the chance simple random sampling: that an unrepresentative element will be ♦♦ It is easier to draw a sample and often used to represent the entire unit. When easier to execute without mistakes. This is a particular advantage when the drawing an observation for an experimental unit is done in the field. is represented as the average of several ♦♦ It ensures that the sample is more spread subsamples, you need to enter only one data across the population. value for the experimental unit. ♦♦ It is more precise. In effect it stratifies the population into a specific number Systematic sample of strata determined from step 1. In a systematic system, the units occur at the In this method, a starting point is selected same relative position in the stratum, at random. Then elements of the sample whereas with the stratified system, the are selected in a systematic, predetermined position in the stratum is determined sequence from the starting point. The three separately by randomization within each steps are: stratum. 1. Divide the number of cases in the ♦♦ It can prevent bias produced by population by the desired sample size. interactions of experimental units 2. in close proximity to each other. Select a random number between one Systematic sampling is sometimes used with pesticides that may drift to other experimental units downwind from the point of application. Disadvantages of systematic sampling: ♦♦ The system is not as precise when periodicity (repeating in a regular order) occurs in the population. However, this problem is easy to determine and the researcher can use various methods—such as stratification—to prevent it. Cluster (area) sample When a population is widely disbursed Figure 26. Systematic sample taken by selecting a over a large geographical area, it can be starting point at random and the sampling every 10th difficult to do random sampling. A simple sampling unit from left to right, wrapping around when necessary. This sampling plan preserves the same random sample may be biased unless the probability of a unit being selected as a simple random population is not uniformly or randomly sample. distributed. If a population is clumped (as in herds of livestock, wild plant communities, or plant diseases), cluster sampling may and the value attained in Step 1. be appropriate. Certain characteristics of 3. Start sampling with the number chosen the population’s structure must be known, in Step 2 and proceed at regular intervals however. There are three steps in cluster indicated by the desired sample size. sampling: Systematic sampling has no subjectivity if 1. Divide the population into clusters the sample location is selected at random (usually along geographic boundaries) 54 | Demonstration and Research Pest Control

2. Randomly sample the clusters. Statistical calculations 3. Measure all units within sampled clusters. To make inferences about the entire population based on the data gathered Cluster sampling is economical, because in the experiment, you must determine if for a given sample size, a small unit often observed differences are truly different or if gives more precise results than a large unit. they are a result of random variation. This is Cluster sampling is efficient because time where statistics comes into play. is not wasted sampling empty or nearly empty units. However, researchers must After collecting data from a properly take precautions because cluster sampling is designed experiment, you will usually susceptible to bias. need to analyze the data with appropriate statistical calculations. Statistical analysis methods are selected based on the experiment’s objectives and design. Proper statistical analysis can be done if your experiment was designed according to the principles outlined in this publication, but proper analysis can be complicated greatly if these principles were not followed. It is probably best to have help in making statistical calculations. Professional statisticians and other scientists may be willing to help you with the statistics if you involve them early in the process (well before you lay out plots). They can also check your proposed design for flaws and omissions. If you want to do the work yourself, some simple statistics can be Figure 28. Cluster (area) sample: 20 samples are calculated by hand; but most people will selected by randomly selecting five clusters of four sampling units from a total of 49 clusters and then make the calculations with the help of sampling all four units in a cluster. computer software. Specialized statistical software is available, but most spreadsheet software can calculate simple statistics. Although a full review of statistical methods Multistage sampling is outside the scope of this manual, many texts are available to help you with this Multistage sampling combines different portion of experimental results analysis. methods of probability sampling (e.g., using cluster sampling to select certain fields, and then doing random sampling within each field). The important principle of multistage sampling is that it can combine simple methods in a variety of useful ways to address sampling needs in the most effective and efficient manner possible. Appendix A - Review questions | 55

Appendix A - Review questions 1. Which persons applying unregistered Which license would be inappropriate experimental pesticides are exempt for applying the experimental pesticide? from the standards of licensing and a. An ODA Consultant license with certification in the demonstration and the Demonstration and Research research categories? category. a. Persons conducting laboratory b. An ODA Commercial Pesticide research and persons with a Ph.D. Applicator License with the degree in entomology. Demonstration and Research b. Persons conducting laboratory Category. research and persons conducting c. An ODA Private Pesticide greenhouse efficacy trials. Applicator license. c. Ph.D.s in entomology, plant d. An ODA Public Pesticide pathology, and weed science, and Applicator License with the M.D.s and D.V.M.s. Demonstration and Research d. None of the above. Category. 2. The experimental application of 5. Who is the responsible party to ensure potential pesticides not registered crop destruction requirements are met? with EPA to more than 10 acres of a. The Private Pesticide Applicator. agricultural land requires: b. The EUP holder. a. Certification in demonstration and research pest control. c. The research technician. b. A Federal Experimental Use Permit d. The licensed Pesticide Consultant. (EUP) issued by EPA. 6. What is the most important regulation c. An Oregon Experimental Use for pesticide researchers submitting data Permit (EUP) issued by ODA. to EPA in support of FIFRA and TSCA actions? d. All of the above. a. Applicator Standard (40 CFR 171). 3. What does an experimental use permit (EUP) allow? b. Good Laboratory Practice Standards (40 CFR 160). a. Limited field testing of an unregistered pesticide or a registered c. Worker Protection Standard (40 pesticide for an unapproved use. CFR 170). b. Limited sales of an unregistered d. Emergency Use Permit requirement pesticide in all states and U.S. (FIFRA sec. 5). territories. 7. If a forage crop has been treated with c. Laboratory testing and greenhouse an experimental pesticide with no screening of chemicals for potential established tolerances, how much time use in agriculture. must elapse before grazing is allowed d. All of these. a. 14 days. 4. A researcher wants to apply a chemical b. 30 days. to three 3-acre plots to test its efficacy c. 60 days. against several pests to determine if it has potential as an agricultural pesticide. d. 365 days. 56 | Demonstration and Research Pest Control

8. When making a pesticide application under an 13. For purposes of GLPS compliance, what do experimental use permit, the permit holder must changes in an approved protocol require? notify ODA at least: a. Telephone call to an EPA regional office. a. 24 hours prior to application. b. A study protocol can not be changed after it is b. 48 hours prior to application. approved. c. 72 hours prior to application. c. Re-submission of the protocol for approval. d. Notification is not required. d. Changes must be documented, signed by the study director, dated, and maintained with the 9. A demonstration differs from a research protocol. experiment in that: 14. Where must specimen information be located, a. Demonstrations require an EUP. such as date of collection, nature and test system, b. Research experiments are not usually to satisfy GLPS compliance? statistically valid. a. On the specimen container or accompanying c. Demonstrations have only registered uses of the specimen. registered products. b. In the protocol. d. Research experiments may only be conducted c. In a secure file cabinet. on OSU research farms. d. In confidential files. 10. To meet GLPS requirements under 40 CFR 160.12, what must all pesticide studies submitted 15. To comply with GLPS, how must all data be to EPA in support of FIFRA sec. 5 EUPs, sec. 3 recorded except data recorded by automated data registrations and 24(c) and sec. 18 uses include? retrieval systems? a. Statement of compliance or non-compliance. a. In ANSI data files. b. Name of trial sponsor. b. In pencil. c. Name of state lead agency. c. In ink. d. All of these. d. In ASCII text files. 11. What can result from failure to comply with 16. If an item of erroneous data was recorded, what applicable GLPS? must accompany the correction to comply with GLPS requirements? a. Cancellation, suspension or modification of the research or marketing permit. a. Telephone call to an EPA regional office. b. Imposition of civil penalties under FIFRA b. A permit to change data from EPA Office of section 14. Pesticides and Toxic Substances (OPTS). c. Criminal prosecution under 18 U.S.C. 2 or c. The reason for the change, date and signature 1001 or FIFRA section 14. of person making the change. d. All of these. d. Changes to data are not permitted under GLPS. 12. For purposes of GLPS compliance, the protocol for a field trial of a pesticide requires which of the 17. Which factor is NOT important in the ability of following? a pesticide to be effective? a. Justification for selection of the test system. a. Penetration. b. A statement of the proposed statistical b. Transport. method to be used. c. Mode of action. c. A description of the experimental design, d. Molecular weight. including methods for the control of bias. d. All of these. Appendix A - Review questions | 57

18. The conversion of 2,4-DB to 2,4-D by beta- 21. The researchers assume that calves will be oxidation in plants is an example of ______. caught “at random.” They enter the compound, catch four calves, and assign them to treatment a. Metabolic deactivation. A. They catch the next four calves and assign b. Mode of action. them to treatment B. They follow this process through treatments C and D. Because the calves c. Metabolic activation. were caught at random, the researchers assume d. Biological magnification. that this constitutes a CRD and they analyze the results accordingly. Which of the following 19. Eggshell thinning of predatory birds induced by statements best characterizes the nature of this DDT is an example of ______. experiment? a. Metabolism. a. The experiment has inadequate replication. b. Activation. b. The experiment has inadequate c. Biodegradation. randomization. d. Biological magnification. c. The experiment is a CRD with serious bias. 20. Your goal is to apply 30 gallons per acre using a d. The experiment is a RCB not a CRD. CO2 backpack sprayer with a 4-nozzle boom and 22. The researchers catch all the calves and label 20-inch nozzle spacing. Your measured walking them 1 to 16. Using a random number table, they speed with the tank half full is 3 miles per hour. select four pens at random to receive the calves How many milliliters should be collected in 30 to which treatments will be assigned. Then, using seconds? (See formula on page 7.) a random number table, they select numbers a. 154.0 ml. corresponding to each numbered calf to receive the first treatment. They place these calves in the b. 271.5 ml. appropriate pen to receive treatment A. Then c. 573.5 ml. they select another four calf numbers at random and put them in the corresponding pen to receive d. 1,147.0 ml. treatment B. They continue until they have four pens with four calves each. Each pen receives Questions 21-30 (experimental design) a different treatment, and the experiment is analyzed as a completely randomized experiment. Researchers want to use a completely randomized Which of the following statements best design (CRD) to compare three different larvicide characterizes the nature of this experiment? treatments formulated in a free-choice mineral mix to a. be fed to feeder calves to control flies. The larvicides The experiment has inadequate replication. are S-methoprene (Altosid®), tetrachlorvinphos b. The experiment has inadequate (Rabon Oral Larvicide® = ROL®), and diflubenzuron randomization. (Vigilante®). This necessitates four treatments, which c. are labeled treatment A (Altosid®), treatment B The experiment is a CRD with serious bias. (ROL®), treatment C (Vigilante®), and treatment D d. The experiment is a RCB not a CRD. (untreated control). The researchers will measure the treatment effects by periodically counting the number of flies on each calf. They will compare the effect of each treatment to the others and to the untreated control. They have 16 calves available for the experiment, so four calves will receive each treatment. The 16 calves arrive and are placed in a large holding compound until they are ready for the experiment. Because flies readily switch from one animal to another, the calves will be transferred to pens at the time of treatment. 58 | Demonstration and Research Pest Control

23. From the following experimental designs, select 24. What is an advantage of an RCB over a CRD the best design for a CRD for the 16 calves to test experimental design? the effects of free-choice oral larvicide mineral a. Permits testing of equality between treatments mixes for fly control. The design should take into by comparing sample variations. account both the variability in calves and the placement of pens. b. Permits the analysis of unbalanced data (unequal replicates or samples). a. Number calves and pens from 1 to 16. Use a random number table to assign a number c. Permits detection of treatment differences in to each calf, which corresponds to a pen the presence of a single extraneous source of number. Use a random number table to variability. assign treatment A to the first four randomly d. Permits detection of differences between selected calves. Repeat this procedure without treatments without replication. replacement until all calves have received a treatment. Place oral larvicides in mineral 25. What is the value of randomization in an feeders in each pen according to the treatment experimental design? number assigned to calves. Then introduce a. It averages out the effect of all unknown, calves according to pen number. Periodically uncontrollable factors in all treatment groups. record the number of flies per animal. b. It makes the experiment powerful enough to b. Number pens 1 to 16, but do not number detect treatment effects in all groups. calves. Put 16 slips of paper in a hat, each numbered 1 to 16. In a second hat, put 16 c. It removes the influence of all unknown, slips of paper, four each of which have been uncontrollable factors in all treatment groups. labeled with one of the four treatments. d. It permits comparisons between all treatment Catch a calf and assign a pen and treatment groups. at random by drawing one slip from the first hat and one slip from the second hat. Repeat 26. Under what condition or conditions would without replacement until all calves have Completely Randomized Designs (CRD) be received a pen and a treatment. Periodically better than Randomized Complete Block (RCB) record the number of flies per animal. designs? c. Place slips of paper in a hat with the letters a. Data is from more than one field or more A, B, C and D printed on four slips each. than one farm. Catch the first calf, pick a slip at random from b. Data is from field and row crop studies rather the hat, and assign the calf to the treatment than livestock feeding studies. letter on the slip. Do not replace the slip. Catch a second calf and select another slip c. Data is unbalanced (replications or sample from the remaining three slips. Assign that numbers differ between treatment groups). treatment to the second calf. Continue until d. Data is balanced (replications or sample the first four calves have receive one of the numbers are equal in all treatment groups). four treatments. Then replace the slips and repeat the process until all 16 calves have a 27. Four different insecticides are to be evaluated treatment. Periodically record the number of for their control of plant bugs in cotton. The flies per animal. data might differ from one location in a field to another because plant bug population d. Number calves from 1 to 16, but do not distributions are highly clumped. Other number pens. Use a random number table to factors such as soil, irrigation and fertility are select the first four calves at random for the comparatively homogenous. Which experimental first treatment. Place these calves in a pen design would be most appropriate? with treatment A. Repeat this procedure until all calves have been caught and placed in a a. Latin square. pen for treatments B through D. Periodically b. CRD. record the number of flies per animal. c. Split plot. d. RCB. Appendix A - Review questions | 59

28. Four different dosage rates of a plant growth regulator are to be evaluated on Pima and upland cotton. The effect will be measured in pounds of lint from each treatment. Which experimental design is most appropriate? a. RCB. b. Latin squares. c. CRD. d. Split plot. 29. Potential phytotoxicity of eight different concentrations of lime-sulfur orchard spray is evaluated on eight blocks of trees containing eight different varieties. Each variety must receive applications of all eight concentrations of chemical. Experimenters want to control for variability from varieties and for variability from the location within the orchard of a particular treatment within a variety. Which experimental design is most appropriate? a. Split plot. b. CRD. c. Latin square. d. RCB. 30. Researchers want to evaluate two herbicide treatments (A and B), each applied at a high and low rate (1 and 2) to be compared with untreated controls (0). Which experimental design is most appropriate? a. RCB. b. Factorial. c. Latin square. d. Split plot. 60 | Demonstration and Research Pest Control Appendix B - Answers to review questions | 61

Appendix B - Answers to review questions each pen are not independent. One calf consuming a lot of the larvicidal 1. d formulation may achieve fly control, 2. d while others in that pen eating less may not get as much fly control. There 3. a is no way to separate the effect of the 4. c high larvicide consumption rate from the lower levels of control in calves 5. b eating less. The experimental unit is the 6. b smallest unit of experimental matter to which the treatment is applied at 7. d random. In this case, the pens are the 8. c experimental units. For a CRD, each calf must have its own pen. 9. c 23. b 10. a For a CRD, label the pens 1 through 11. d 16. Do not number calves. Place 16 slips of paper numbered from 1 to 16 12. d in a hat. In a second hat, place 16 slips 13. d of paper corresponding to treatments, four each labeled A, B, C and D. Catch 14. a a calf. Select a number and a letter from 15. c each hat. Place the calf in the location indicated by the number and feed it the 16. c treatment assigned by the letter. Repeat 17. d without replacement until all calves have been assigned a treatment and pen. 18. c 24. c 19. d It permits detection of treatment 20. dc differences in the presence of a single extraneous source of variability. The 21. b RCB design improves precision (relative The experiment has inadequate to CRD) by enabling the experimenter randomization. The first calves caught to detect and control a single extraneous, could be the slowest, weakest, least non-random source of variation between physically active, least able to escape and among experimental units and to capture, and possibly least inclined to remove its effect from the estimate of use physical activity to avoid annoyance experimental error. This increases scope by flies. This would bias the results. If of inference and allows the experimenter the experimental results came out to to make broader conclusions. the disadvantage of treatment A, there would be no way to determine if the 25. a results were a consequence of treatment It averages out the effects of all A or the fact that the weakest calves unknown, uncontrollable factors in were placed on that treatment by the all treatment groups. Randomization selection process. is used to control random variability among individual experimental units. 22. a The experiment is a seriously flawed 26. c CRD. It completely lacks replication. Data is unbalanced (replications There are 16 calves, but the calves in or sample numbers differ between 62 | Demonstration and Research Pest Control

treatment groups). The advantages of of variables, or factors, on some response. the completely randomized design are There are advantages to combining that it is flexible and simple. It is the the study of several factors in the same only design that permits the analysis factorial experiment. The treatments of unbalanced data, which consists of consist of all combinations that can be unequal numbers of replications or formed from the different factors. Each samples between treatment groups. treatment combination is randomized This is particularly advantageous when within each replication. Such a design weather or other factors destroy one is useful when studying the effects of or more of the replications in the several rates of several pesticides applied experiment. in the same experiment. 27. d When a single source of natural variability exists within an otherwise homogenous field, in this case the clumped distribution of plant bugs, a randomized complete block (RCB) design is most appropriate. 28. d A split plot design is most appropriate when experiments evaluate both pesticide performance and crop management practices (in this case two different species of cotton that may react differently to varying treatment rates of a plant growth regulator), because it controls for the variability between Pima and upland cotton. It also is applicable in the performance of different cotton varieties. 29. c A Latin square design is most appropriate when evaluating the effects of different treatment rates of the same pesticide on an equal number of experimental units. It is very useful in controlling experimental error from two sources, in this case tree varieties and their treatment location within the orchard. It groups treatments in two different ways—by columns and rows (Fig. 17). Every treatment occurs once in each block (row = tree variety) and once in each column (treatment rate). Variability across the experimental area is measured and removed in two directions. With the Latin square design, the number of treatments must equal the number of replications. With a large number of treatments this design becomes cumbersome. Usually, this design is used for experiments where there are four to eight treatments. 30. b A factorial design is most appropriate when investigating the effects and possible interaction of each of a number Appendix C - Glossary | 63

Appendix C - Glossary

Acetolactate synthase (ALS) inhibitors Chloroplast Group of herbicides that inhibit an enzyme Organelles present in large numbers within that is required in the process of forming plant cells; they contain protein, lipids and several essential plant amino acids. pigments. Acetylcholine Cholinesterase

An enzyme that transmits nerve signals The enzyme that destroys acetylcholine. between nerves and muscles, sensory organs, or other nerves. Colloidal fraction Amino acids That portion of a soil consisting of clay and organic matter, in which most of the surface Basic building blocks of proteins in plants reactions take place. and animals. Dithiocarbamate fungicide Antagonism Group of sulfur-containing, broad-spectrum, Reduced toxicity or effectiveness as a result synthetic organic contact fungicides used to of combining one pesticide with another. prevent many fungal diseases. Antidote. Substance that detoxifies the effects of a particular pesticide. Enzyme Biodegradation Proteins, formed in plant and animal cells or made synthetically, that act as organic The decomposition of pesticide residues catalysts in initiating or speeding up specific in the environment by bacteria and other chemical reactions. microorganisms that use the residues as nutrient sources. Fumigant Biological magnification Liquid or solid chemical that forms a gas and kills organisms. The tendency of certain pesticides to become progressively more concentrated in Fungicide each type of organism when moving from the bottom to the top organism in a food A pesticide used to prevent, repel or chain. mitigate fungal infections. Calibration Growth regulator

The process of measuring the output of A hormonal chemical that changes the pesticide application equipment so that the growth of a plant or animal. proper amount of pesticide can be applied to Herbicide a given area. Chloroacetamide herbicide A pesticide used to prevent or destroy plants that grow where they are not wanted. Family of herbicides, including acetochlor, Insecticide alachlor, metolachlor and propachlor; they are most commonly soil-applied and inhibit A pesticide used to prevent, destroy, repel, emerging roots and shoots. mitigate or attract insects and their relatives. 64 | Demonstration and Research Pest Control

Ionization The biological production of organic substances, chiefly sugars, in green plant Process in which a chemical takes on a cells in the presence of light. positive or negative electrical charge because of losing or gaining electrons through a Phytotoxicity chemical reaction. An effect that is injurious or lethal to plants. Metabolism Postemergence herbicide The total chemical process that takes place in a living organism to utilize food and A herbicide applied after emergence of the manage wastes, grow and reproduce, and specified weed or crop. accomplish all other life functions. Pre-emergence herbicide Metabolite A herbicide applied to the soil before A substance produced in a living organism emergence of the specified weed or crop. through metabolism. Resistance Mode of action Genetic qualities of some individuals in a The mechanism(s) by which pesticides pest population that enable them to resist injure or kill pests. the effects of certain types of pesticides that are toxic to other members of that species. Nematicide Restricted-use pesticide A pesticide used to prevent, repel, destroy or mitigate the effects of nematodes. A pesticide that can be sold only to or applied only by individuals licensed as Organic matter commercial or private applicators, or persons under their direct supervision. Portion of soil containing carbon and hydrogen. Selectivity

Organochlorine compound The ability of a pesticide to kill some pests but not others. A class of pesticides, commonly used as insecticides, that contain a chlorine atom Soil texture incorporated into an organic molecule. Organochlorines are often highly persistent. Relative content percentages of clay, sand and silt in any given soil. Organophosphate (OP) Synergism A commonly used class of pesticides containing various esters of phosphoric and The action of two pesticides that produces a related acids, which break down rapidly in greater cumulative effect when the pesticides the environment. are used together than when they are used individually. Permeability Systemic herbicide Referring to a surface or membrane that is porous, allowing certain substances to pass A herbicide that is absorbed by treated through. plants and translocated to other tissues by the vascular system. Pesticide

A chemical or mixture of chemicals used to destroy, prevent, repel or attract any animal, plant, or disease considered a pest. Photosynthesis Appendix D - Table of random numbers | 65 Appendix D - Table of random numbers To randomize any set of items, begin at a random point on the table and follow rows or columns or diagonals in either direction. Write down the numbers in the order they appear. Disregard those that are higher than the number being randomized and those that have appeared before in the series. 1 2 3 4 5 6 7 8 9 10 1 21911 57448 71407 85284 01680 62604 28599 97996 03365 46822 2 08933 85456 66308 63725 22823 71580 22662 53885 60806 82919 3 92813 81352 10192 86313 35178 79693 30003 46966 49823 10409 4 78135 36897 34431 85533 56513 21232 77743 20413 33244 79255 5 55593 73656 86167 62292 42919 71584 09918 99489 50735 97693 6 42570 14906 81806 90913 02421 47320 41347 98838 84270 20637 7 01615 42647 97796 30519 92483 40326 89758 91764 77973 86128 8 76810 18089 75569 18262 76648 21511 97692 61378 68959 79185 9 45762 57872 85805 89330 89772 91828 47231 58712 07490 63458 10 14603 48631 18296 68570 05162 99434 65418 38800 64304 35037 11 16629 48215 01753 93813 55932 04523 94947 66732 74567 78866 12 44738 20178 75907 37312 45505 84754 60507 82793 83726 10232 13 88155 45115 29334 58055 00899 39706 54052 43383 11091 30342 14 89131 40800 03158 15118 87746 26363 43430 69032 04938 85554 15 88399 63167 49089 78284 84267 02267 53133 52514 25266 21315 16 36929 47098 67526 35752 08624 22590 74969 59467 44664 18934 17 07322 15133 66923 90027 06624 14655 48772 92192 60054 24591 18 73903 44704 01127 32984 38692 62108 91794 89562 94933 91407 19 17913 41059 16976 80914 81066 94937 55820 84148 62285 39948 20 76811 76369 85182 44049 65443 36205 58074 48824 06606 25620 21 20112 03774 77759 41096 75639 23585 26597 31248 40448 65570 22 58937 65881 67280 89861 96876 27158 00875 02925 29904 61577 23 22534 88578 24653 45976 21794 10660 15915 53580 99835 60275 24 81234 85761 78984 77538 28255 68462 39696 03273 81152 20528 25 36760 56181 31608 63751 94352 81634 48323 81910 41845 03355 26 26177 60957 88131 28522 61075 88383 61094 47139 06733 09061 27 73152 15694 63995 17888 01301 32195 61157 45832 03384 58168 28 81945 27477 99059 23516 97786 78017 01784 66843 64859 43563 29 29863 23030 32979 45036 76467 42070 78270 80389 82792 86593 30 34591 50367 37701 66337 01661 45129 00757 28640 21402 14548 31 44641 06719 88077 52531 43111 66356 68701 47138 11346 89469 32 81374 44328 11474 11135 03039 82172 70424 94459 23263 13791 33 74605 99521 11685 61134 01907 14968 50552 19646 39264 11228 34 83193 76961 96491 56585 00814 40277 60630 33310 41892 09529 35 60991 12296 40892 06576 74477 68394 14659 93724 91639 18625 36 47129 10311 94296 07605 40038 42207 22556 35289 07661 30061 37 54480 72438 69950 18716 71093 82524 89346 46641 05849 86806 38 13204 42878 05262 83601 73323 17580 28239 03257 26386 10131 39 85223 12355 79622 94628 28360 67254 29231 35885 08003 26625 40 95537 51318 38757 36118 83581 74390 68185 74924 32863 79001 66 | Demonstration and Research Pest Control Appendix E - Latin square design | 67

Appendix E - Latin square design

Table of Latin squares design The following are examples of Latin squares design. Computer programs are available to help create other Latin squares. One such program can be found here: http://www.jaapsch.net/sudoku.htm 2 5 2 A B B A

3 53 A B C B C A C A B

4 54 A B C D A B C D A B C D A B C D B C D A B D A C B A D C B A D C C D A B C A D B C D A B C D B A D A B C D C B A D C B A D C A B

5 5 5 A B C D E B D E A C C A D E B D E B C A E C A B D

6 5 6 A B C D E F B F D C A E C A E B F D D E B F C A E C F A D B F D A E B C

7 5 7 A B C D E F G B E G A C D F C A D B F G E D F A E G C B E C F G B A D F G B C D E A G D E F A B C 68 | Demonstration and Research Pest Control

85 8 A B C D E F G H B D A H G C E F C E G F A B H D D A B C H G F E E C H A F D B G F G D E B H A C G H F B C E D A H F E G D A C B

95 9 A B C D E F G H I B H F E C A I D G C A H I D B E G F D I B C G E A F H E C I F B G H A D F G E A H I D B C G D A B I H F C E H E D G F C B I A I F G H A D C E B

10 510 A B C D E F G H I J B F J I A C E D G H C A G E D J I B H F D H I A C E B J F G E C A F J G H I B D F D E B G H A C J I G J D C H I F A E B H I B G F A J E D C I G H J B D C F A E J E F H I B D G C A

Appendix F - Pesticide contacts

Appendix F - Pesticide contacts

Pesticide emergency information The National Pesticide Information Center (NPIC) Oregon Poison Control Center 1-800-858-7378 1-800-222-1222 The National Pesticide Information Center (NPIC) The Oregon Poison Control Center provides can provide information about pesticide products and emergency treatment information for cases of their toxicity. poisonings or toxic exposures. Access to this health care advice is available to both the public and health U.S. Environmental Protection Agency care providers in the state. This is a 24-hour toll- free resource that provides access to a network of (EPA), Region VI Spill Reporting nurses, pharmacists, paramedics and physicians who have extensive education, training and expertise 1-800-832-8224 in the field of toxicology. The Poison Control Center also provides public education activities for All major pesticide spills must by law be reported teachers, students and citizens, as well as educational immediately to the U.S. Environmental Protection opportunities for Oregon health care providers. Agency, Region 10 Office, 1200 Sixth Ave Suite 900, Seattle, WA, by calling 1-800-424-4EPA or Oregon Emergency Response Center 206-253-1250. EPA Region 10 emergency response to chemical spills may be accessed 24 hours a day 1-800-452-0311 at the Region VI Emergency Response Center: 1-866-EPASPILL (866-372-7745). The following OERS was established in 1972 by the Governor information should be reported: of Oregon to improve communications and 1. Name, address and telephone number of person coordination between government agencies reporting responding to hazardous materials incidents 2. (including pesticides) across the state. Exact location of spill 3. Since that time OERS has become an “all-hazards” Name of company involved and location system, responding to other types of emergencies: 4. Specific pesticide spilled Natural Hazards such as floods, wildfire and 5. earthquakes and Search and Rescue missions. Estimated quantity of pesticide spilled 6. OERS operates under Oregon Revised Statutes Source of spill (ORS) 401, Executive Order of the Governor and 7. Cause of spill Oregon Administrative Rules Chapter 104, Division 8. 40. Name of body of water involved, or nearest body of water to the spill area National Pesticide Safety Team Network 9. Action taken for containment and cleanup (Chemtrec) 1-800-424-9300 The National Agricultural Chemicals Association has a telephone network. This network can tell the applicator the correct contamination procedures to use to send a local safety team to clean up a spill. An applicator can call the network toll free at any time.